diff --git "a/data/multi_agent/full.jsonl" "b/data/multi_agent/full.jsonl"
new file mode 100644--- /dev/null
+++ "b/data/multi_agent/full.jsonl"
@@ -0,0 +1,360 @@
+{"id": "cloud_deployment_0240_closed_missing_direct__cwa__multi_agent", "source_id": "cloud_deployment_0240_closed_missing_direct__cwa", "base_id": "cloud_deployment_0240_closed_missing_direct", "split": "train", "domain": "cloud_deployment", "family": "closed_missing_direct", "subset": "core_contrastive", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["has_deployment_approval", "has_incident_note", "may_deploy", "needs_sre_review", "passed_unit_tests"], "agent_a_complete_predicates": ["has_deployment_approval", "has_incident_note", "may_deploy", "needs_sre_review", "passed_unit_tests"], "agent_b_open_predicates": [], "facts_positive": ["Service A passed unit tests."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_unit_tests": "passed unit tests", "has_deployment_approval": "has deployment approval", "has_incident_note": "has an incident note", "may_deploy": "may deploy to production", "needs_sre_review": "needs SRE review"}, "vocabulary_predicates": ["has_deployment_approval", "has_incident_note", "may_deploy", "needs_sre_review", "passed_unit_tests"], "query_atom": "has_deployment_approval::Service A", "target_statement": "Service A has deployment approval.", "gold_reason_type": "global_closed_world_underivable", "symbolic": {"positive_atoms": ["passed_unit_tests::Service A"], "negative_atoms": [], "rules": [], "query_atom": "has_deployment_approval::Service A", "closure_atoms": ["passed_unit_tests::Service A"]}, "source_reports": {"agent_a": {"positive": ["Service A passed unit tests."], "negative": []}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "closed_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_deployment_approval, has_incident_note, may_deploy, needs_sre_review, passed_unit_tests.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- has_deployment_approval: has deployment approval\n- has_incident_note: has an incident note\n- may_deploy: may deploy to production\n- needs_sre_review: needs SRE review\n- passed_unit_tests: passed unit tests\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- Service A passed unit tests.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: has_deployment_approval::Service A.\nTarget statement: Service A has deployment approval.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "cloud_deployment_0240_closed_missing_direct__lcwa__multi_agent", "source_id": "cloud_deployment_0240_closed_missing_direct__lcwa", "base_id": "cloud_deployment_0240_closed_missing_direct", "split": "train", "domain": "cloud_deployment", "family": "closed_missing_direct", "subset": "core_contrastive", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_deployment_approval"], "agent_a_complete_predicates": ["has_deployment_approval"], "agent_b_open_predicates": ["has_incident_note", "may_deploy", "needs_sre_review", "passed_unit_tests"], "facts_positive": ["Service A passed unit tests."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_unit_tests": "passed unit tests", "has_deployment_approval": "has deployment approval", "has_incident_note": "has an incident note", "may_deploy": "may deploy to production", "needs_sre_review": "needs SRE review"}, "vocabulary_predicates": ["has_deployment_approval", "has_incident_note", "may_deploy", "needs_sre_review", "passed_unit_tests"], "query_atom": "has_deployment_approval::Service A", "target_statement": "Service A has deployment approval.", "gold_reason_type": "local_closed_world_underivable", "symbolic": {"positive_atoms": ["passed_unit_tests::Service A"], "negative_atoms": [], "rules": [], "query_atom": "has_deployment_approval::Service A", "closure_atoms": ["passed_unit_tests::Service A"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Service A passed unit tests."], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "closed_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_deployment_approval.\nAgent B open/incomplete predicates: has_incident_note, may_deploy, needs_sre_review, passed_unit_tests.\n\nPredicate glossary:\n- has_deployment_approval: has deployment approval\n- has_incident_note: has an incident note\n- may_deploy: may deploy to production\n- needs_sre_review: needs SRE review\n- passed_unit_tests: passed unit tests\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Service A passed unit tests.\nNegative facts:\n- none\n\nTarget atom: has_deployment_approval::Service A.\nTarget statement: Service A has deployment approval.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "cloud_deployment_0240_closed_missing_direct__owa__multi_agent", "source_id": "cloud_deployment_0240_closed_missing_direct__owa", "base_id": "cloud_deployment_0240_closed_missing_direct", "split": "train", "domain": "cloud_deployment", "family": "closed_missing_direct", "subset": "core_contrastive", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["has_deployment_approval", "has_incident_note", "may_deploy", "needs_sre_review", "passed_unit_tests"], "facts_positive": ["Service A passed unit tests."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_unit_tests": "passed unit tests", "has_deployment_approval": "has deployment approval", "has_incident_note": "has an incident note", "may_deploy": "may deploy to production", "needs_sre_review": "needs SRE review"}, "vocabulary_predicates": ["has_deployment_approval", "has_incident_note", "may_deploy", "needs_sre_review", "passed_unit_tests"], "query_atom": "has_deployment_approval::Service A", "target_statement": "Service A has deployment approval.", "gold_reason_type": "open_world_underivable", "symbolic": {"positive_atoms": ["passed_unit_tests::Service A"], "negative_atoms": [], "rules": [], "query_atom": "has_deployment_approval::Service A", "closure_atoms": ["passed_unit_tests::Service A"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Service A passed unit tests."], "negative": []}}, "gold_truth_value": "unknown", "gold_source_used": "agent_b", "gold_closure_handling": "open_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: has_deployment_approval, has_incident_note, may_deploy, needs_sre_review, passed_unit_tests.\n\nPredicate glossary:\n- has_deployment_approval: has deployment approval\n- has_incident_note: has an incident note\n- may_deploy: may deploy to production\n- needs_sre_review: needs SRE review\n- passed_unit_tests: passed unit tests\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Service A passed unit tests.\nNegative facts:\n- none\n\nTarget atom: has_deployment_approval::Service A.\nTarget statement: Service A has deployment approval.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "cloud_deployment_0241_closed_missing_direct__cwa__multi_agent", "source_id": "cloud_deployment_0241_closed_missing_direct__cwa", "base_id": "cloud_deployment_0241_closed_missing_direct", "split": "train", "domain": "cloud_deployment", "family": "closed_missing_direct", "subset": "core_contrastive", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["has_deployment_approval", "has_incident_note", "may_deploy", "needs_sre_review", "passed_unit_tests"], "agent_a_complete_predicates": ["has_deployment_approval", "has_incident_note", "may_deploy", "needs_sre_review", "passed_unit_tests"], "agent_b_open_predicates": [], "facts_positive": ["Service B passed unit tests."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_unit_tests": "passed unit tests", "has_deployment_approval": "has deployment approval", "has_incident_note": "has an incident note", "may_deploy": "may deploy to production", "needs_sre_review": "needs SRE review"}, "vocabulary_predicates": ["has_deployment_approval", "has_incident_note", "may_deploy", "needs_sre_review", "passed_unit_tests"], "query_atom": "has_deployment_approval::Service B", "target_statement": "Service B has deployment approval.", "gold_reason_type": "global_closed_world_underivable", "symbolic": {"positive_atoms": ["passed_unit_tests::Service B"], "negative_atoms": [], "rules": [], "query_atom": "has_deployment_approval::Service B", "closure_atoms": ["passed_unit_tests::Service B"]}, "source_reports": {"agent_a": {"positive": ["Service B passed unit tests."], "negative": []}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "closed_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_deployment_approval, has_incident_note, may_deploy, needs_sre_review, passed_unit_tests.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- has_deployment_approval: has deployment approval\n- has_incident_note: has an incident note\n- may_deploy: may deploy to production\n- needs_sre_review: needs SRE review\n- passed_unit_tests: passed unit tests\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- Service B passed unit tests.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: has_deployment_approval::Service B.\nTarget statement: Service B has deployment approval.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "cloud_deployment_0241_closed_missing_direct__lcwa__multi_agent", "source_id": "cloud_deployment_0241_closed_missing_direct__lcwa", "base_id": "cloud_deployment_0241_closed_missing_direct", "split": "train", "domain": "cloud_deployment", "family": "closed_missing_direct", "subset": "core_contrastive", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_deployment_approval"], "agent_a_complete_predicates": ["has_deployment_approval"], "agent_b_open_predicates": ["has_incident_note", "may_deploy", "needs_sre_review", "passed_unit_tests"], "facts_positive": ["Service B passed unit tests."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_unit_tests": "passed unit tests", "has_deployment_approval": "has deployment approval", "has_incident_note": "has an incident note", "may_deploy": "may deploy to production", "needs_sre_review": "needs SRE review"}, "vocabulary_predicates": ["has_deployment_approval", "has_incident_note", "may_deploy", "needs_sre_review", "passed_unit_tests"], "query_atom": "has_deployment_approval::Service B", "target_statement": "Service B has deployment approval.", "gold_reason_type": "local_closed_world_underivable", "symbolic": {"positive_atoms": ["passed_unit_tests::Service B"], "negative_atoms": [], "rules": [], "query_atom": "has_deployment_approval::Service B", "closure_atoms": ["passed_unit_tests::Service B"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Service B passed unit tests."], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "closed_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_deployment_approval.\nAgent B open/incomplete predicates: has_incident_note, may_deploy, needs_sre_review, passed_unit_tests.\n\nPredicate glossary:\n- has_deployment_approval: has deployment approval\n- has_incident_note: has an incident note\n- may_deploy: may deploy to production\n- needs_sre_review: needs SRE review\n- passed_unit_tests: passed unit tests\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Service B passed unit tests.\nNegative facts:\n- none\n\nTarget atom: has_deployment_approval::Service B.\nTarget statement: Service B has deployment approval.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "cloud_deployment_0241_closed_missing_direct__owa__multi_agent", "source_id": "cloud_deployment_0241_closed_missing_direct__owa", "base_id": "cloud_deployment_0241_closed_missing_direct", "split": "train", "domain": "cloud_deployment", "family": "closed_missing_direct", "subset": "core_contrastive", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["has_deployment_approval", "has_incident_note", "may_deploy", "needs_sre_review", "passed_unit_tests"], "facts_positive": ["Service B passed unit tests."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_unit_tests": "passed unit tests", "has_deployment_approval": "has deployment approval", "has_incident_note": "has an incident note", "may_deploy": "may deploy to production", "needs_sre_review": "needs SRE review"}, "vocabulary_predicates": ["has_deployment_approval", "has_incident_note", "may_deploy", "needs_sre_review", "passed_unit_tests"], "query_atom": "has_deployment_approval::Service B", "target_statement": "Service B has deployment approval.", "gold_reason_type": "open_world_underivable", "symbolic": {"positive_atoms": ["passed_unit_tests::Service B"], "negative_atoms": [], "rules": [], "query_atom": "has_deployment_approval::Service B", "closure_atoms": ["passed_unit_tests::Service B"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Service B passed unit tests."], "negative": []}}, "gold_truth_value": "unknown", "gold_source_used": "agent_b", "gold_closure_handling": "open_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: has_deployment_approval, has_incident_note, may_deploy, needs_sre_review, passed_unit_tests.\n\nPredicate glossary:\n- has_deployment_approval: has deployment approval\n- has_incident_note: has an incident note\n- may_deploy: may deploy to production\n- needs_sre_review: needs SRE review\n- passed_unit_tests: passed unit tests\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Service B passed unit tests.\nNegative facts:\n- none\n\nTarget atom: has_deployment_approval::Service B.\nTarget statement: Service B has deployment approval.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "cloud_deployment_0243_closed_missing_direct__cwa__multi_agent", "source_id": "cloud_deployment_0243_closed_missing_direct__cwa", "base_id": "cloud_deployment_0243_closed_missing_direct", "split": "train", "domain": "cloud_deployment", "family": "closed_missing_direct", "subset": "core_contrastive", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["has_deployment_approval", "has_incident_note", "may_deploy", "needs_sre_review", "passed_unit_tests"], "agent_a_complete_predicates": ["has_deployment_approval", "has_incident_note", "may_deploy", "needs_sre_review", "passed_unit_tests"], "agent_b_open_predicates": [], "facts_positive": ["Service D passed unit tests."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_unit_tests": "passed unit tests", "has_deployment_approval": "has deployment approval", "has_incident_note": "has an incident note", "may_deploy": "may deploy to production", "needs_sre_review": "needs SRE review"}, "vocabulary_predicates": ["has_deployment_approval", "has_incident_note", "may_deploy", "needs_sre_review", "passed_unit_tests"], "query_atom": "has_deployment_approval::Service D", "target_statement": "Service D has deployment approval.", "gold_reason_type": "global_closed_world_underivable", "symbolic": {"positive_atoms": ["passed_unit_tests::Service D"], "negative_atoms": [], "rules": [], "query_atom": "has_deployment_approval::Service D", "closure_atoms": ["passed_unit_tests::Service D"]}, "source_reports": {"agent_a": {"positive": ["Service D passed unit tests."], "negative": []}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "closed_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_deployment_approval, has_incident_note, may_deploy, needs_sre_review, passed_unit_tests.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- has_deployment_approval: has deployment approval\n- has_incident_note: has an incident note\n- may_deploy: may deploy to production\n- needs_sre_review: needs SRE review\n- passed_unit_tests: passed unit tests\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- Service D passed unit tests.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: has_deployment_approval::Service D.\nTarget statement: Service D has deployment approval.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "cloud_deployment_0243_closed_missing_direct__lcwa__multi_agent", "source_id": "cloud_deployment_0243_closed_missing_direct__lcwa", "base_id": "cloud_deployment_0243_closed_missing_direct", "split": "train", "domain": "cloud_deployment", "family": "closed_missing_direct", "subset": "core_contrastive", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_deployment_approval"], "agent_a_complete_predicates": ["has_deployment_approval"], "agent_b_open_predicates": ["has_incident_note", "may_deploy", "needs_sre_review", "passed_unit_tests"], "facts_positive": ["Service D passed unit tests."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_unit_tests": "passed unit tests", "has_deployment_approval": "has deployment approval", "has_incident_note": "has an incident note", "may_deploy": "may deploy to production", "needs_sre_review": "needs SRE review"}, "vocabulary_predicates": ["has_deployment_approval", "has_incident_note", "may_deploy", "needs_sre_review", "passed_unit_tests"], "query_atom": "has_deployment_approval::Service D", "target_statement": "Service D has deployment approval.", "gold_reason_type": "local_closed_world_underivable", "symbolic": {"positive_atoms": ["passed_unit_tests::Service D"], "negative_atoms": [], "rules": [], "query_atom": "has_deployment_approval::Service D", "closure_atoms": ["passed_unit_tests::Service D"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Service D passed unit tests."], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "closed_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_deployment_approval.\nAgent B open/incomplete predicates: has_incident_note, may_deploy, needs_sre_review, passed_unit_tests.\n\nPredicate glossary:\n- has_deployment_approval: has deployment approval\n- has_incident_note: has an incident note\n- may_deploy: may deploy to production\n- needs_sre_review: needs SRE review\n- passed_unit_tests: passed unit tests\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Service D passed unit tests.\nNegative facts:\n- none\n\nTarget atom: has_deployment_approval::Service D.\nTarget statement: Service D has deployment approval.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "cloud_deployment_0243_closed_missing_direct__owa__multi_agent", "source_id": "cloud_deployment_0243_closed_missing_direct__owa", "base_id": "cloud_deployment_0243_closed_missing_direct", "split": "train", "domain": "cloud_deployment", "family": "closed_missing_direct", "subset": "core_contrastive", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["has_deployment_approval", "has_incident_note", "may_deploy", "needs_sre_review", "passed_unit_tests"], "facts_positive": ["Service D passed unit tests."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_unit_tests": "passed unit tests", "has_deployment_approval": "has deployment approval", "has_incident_note": "has an incident note", "may_deploy": "may deploy to production", "needs_sre_review": "needs SRE review"}, "vocabulary_predicates": ["has_deployment_approval", "has_incident_note", "may_deploy", "needs_sre_review", "passed_unit_tests"], "query_atom": "has_deployment_approval::Service D", "target_statement": "Service D has deployment approval.", "gold_reason_type": "open_world_underivable", "symbolic": {"positive_atoms": ["passed_unit_tests::Service D"], "negative_atoms": [], "rules": [], "query_atom": "has_deployment_approval::Service D", "closure_atoms": ["passed_unit_tests::Service D"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Service D passed unit tests."], "negative": []}}, "gold_truth_value": "unknown", "gold_source_used": "agent_b", "gold_closure_handling": "open_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: has_deployment_approval, has_incident_note, may_deploy, needs_sre_review, passed_unit_tests.\n\nPredicate glossary:\n- has_deployment_approval: has deployment approval\n- has_incident_note: has an incident note\n- may_deploy: may deploy to production\n- needs_sre_review: needs SRE review\n- passed_unit_tests: passed unit tests\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Service D passed unit tests.\nNegative facts:\n- none\n\nTarget atom: has_deployment_approval::Service D.\nTarget statement: Service D has deployment approval.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "cloud_deployment_0244_open_missing_direct__cwa__multi_agent", "source_id": "cloud_deployment_0244_open_missing_direct__cwa", "base_id": "cloud_deployment_0244_open_missing_direct", "split": "train", "domain": "cloud_deployment", "family": "open_missing_direct", "subset": "core_contrastive", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["has_deployment_approval", "has_incident_note", "may_deploy", "needs_sre_review", "passed_unit_tests"], "agent_a_complete_predicates": ["has_deployment_approval", "has_incident_note", "may_deploy", "needs_sre_review", "passed_unit_tests"], "agent_b_open_predicates": [], "facts_positive": ["Service E passed unit tests."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_unit_tests": "passed unit tests", "has_deployment_approval": "has deployment approval", "has_incident_note": "has an incident note", "may_deploy": "may deploy to production", "needs_sre_review": "needs SRE review"}, "vocabulary_predicates": ["has_deployment_approval", "has_incident_note", "may_deploy", "needs_sre_review", "passed_unit_tests"], "query_atom": "has_incident_note::Service E", "target_statement": "Service E has an incident note.", "gold_reason_type": "global_closed_world_underivable", "symbolic": {"positive_atoms": ["passed_unit_tests::Service E"], "negative_atoms": [], "rules": [], "query_atom": "has_incident_note::Service E", "closure_atoms": ["passed_unit_tests::Service E"]}, "source_reports": {"agent_a": {"positive": ["Service E passed unit tests."], "negative": []}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "closed_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_deployment_approval, has_incident_note, may_deploy, needs_sre_review, passed_unit_tests.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- has_deployment_approval: has deployment approval\n- has_incident_note: has an incident note\n- may_deploy: may deploy to production\n- needs_sre_review: needs SRE review\n- passed_unit_tests: passed unit tests\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- Service E passed unit tests.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: has_incident_note::Service E.\nTarget statement: Service E has an incident note.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "cloud_deployment_0244_open_missing_direct__lcwa__multi_agent", "source_id": "cloud_deployment_0244_open_missing_direct__lcwa", "base_id": "cloud_deployment_0244_open_missing_direct", "split": "train", "domain": "cloud_deployment", "family": "open_missing_direct", "subset": "core_contrastive", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_deployment_approval", "may_deploy"], "agent_a_complete_predicates": ["has_deployment_approval", "may_deploy"], "agent_b_open_predicates": ["has_incident_note", "needs_sre_review", "passed_unit_tests"], "facts_positive": ["Service E passed unit tests."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_unit_tests": "passed unit tests", "has_deployment_approval": "has deployment approval", "has_incident_note": "has an incident note", "may_deploy": "may deploy to production", "needs_sre_review": "needs SRE review"}, "vocabulary_predicates": ["has_deployment_approval", "has_incident_note", "may_deploy", "needs_sre_review", "passed_unit_tests"], "query_atom": "has_incident_note::Service E", "target_statement": "Service E has an incident note.", "gold_reason_type": "local_open_world_underivable", "symbolic": {"positive_atoms": ["passed_unit_tests::Service E"], "negative_atoms": [], "rules": [], "query_atom": "has_incident_note::Service E", "closure_atoms": ["passed_unit_tests::Service E"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Service E passed unit tests."], "negative": []}}, "gold_truth_value": "unknown", "gold_source_used": "agent_b", "gold_closure_handling": "open_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_deployment_approval, may_deploy.\nAgent B open/incomplete predicates: has_incident_note, needs_sre_review, passed_unit_tests.\n\nPredicate glossary:\n- has_deployment_approval: has deployment approval\n- has_incident_note: has an incident note\n- may_deploy: may deploy to production\n- needs_sre_review: needs SRE review\n- passed_unit_tests: passed unit tests\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Service E passed unit tests.\nNegative facts:\n- none\n\nTarget atom: has_incident_note::Service E.\nTarget statement: Service E has an incident note.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "cloud_deployment_0244_open_missing_direct__owa__multi_agent", "source_id": "cloud_deployment_0244_open_missing_direct__owa", "base_id": "cloud_deployment_0244_open_missing_direct", "split": "train", "domain": "cloud_deployment", "family": "open_missing_direct", "subset": "core_contrastive", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["has_deployment_approval", "has_incident_note", "may_deploy", "needs_sre_review", "passed_unit_tests"], "facts_positive": ["Service E passed unit tests."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_unit_tests": "passed unit tests", "has_deployment_approval": "has deployment approval", "has_incident_note": "has an incident note", "may_deploy": "may deploy to production", "needs_sre_review": "needs SRE review"}, "vocabulary_predicates": ["has_deployment_approval", "has_incident_note", "may_deploy", "needs_sre_review", "passed_unit_tests"], "query_atom": "has_incident_note::Service E", "target_statement": "Service E has an incident note.", "gold_reason_type": "open_world_underivable", "symbolic": {"positive_atoms": ["passed_unit_tests::Service E"], "negative_atoms": [], "rules": [], "query_atom": "has_incident_note::Service E", "closure_atoms": ["passed_unit_tests::Service E"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Service E passed unit tests."], "negative": []}}, "gold_truth_value": "unknown", "gold_source_used": "agent_b", "gold_closure_handling": "open_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: has_deployment_approval, has_incident_note, may_deploy, needs_sre_review, passed_unit_tests.\n\nPredicate glossary:\n- has_deployment_approval: has deployment approval\n- has_incident_note: has an incident note\n- may_deploy: may deploy to production\n- needs_sre_review: needs SRE review\n- passed_unit_tests: passed unit tests\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Service E passed unit tests.\nNegative facts:\n- none\n\nTarget atom: has_incident_note::Service E.\nTarget statement: Service E has an incident note.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "cloud_deployment_0246_open_missing_direct__cwa__multi_agent", "source_id": "cloud_deployment_0246_open_missing_direct__cwa", "base_id": "cloud_deployment_0246_open_missing_direct", "split": "train", "domain": "cloud_deployment", "family": "open_missing_direct", "subset": "core_contrastive", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["has_deployment_approval", "has_incident_note", "may_deploy", "needs_sre_review", "passed_unit_tests"], "agent_a_complete_predicates": ["has_deployment_approval", "has_incident_note", "may_deploy", "needs_sre_review", "passed_unit_tests"], "agent_b_open_predicates": [], "facts_positive": ["Service A passed unit tests."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_unit_tests": "passed unit tests", "has_deployment_approval": "has deployment approval", "has_incident_note": "has an incident note", "may_deploy": "may deploy to production", "needs_sre_review": "needs SRE review"}, "vocabulary_predicates": ["has_deployment_approval", "has_incident_note", "may_deploy", "needs_sre_review", "passed_unit_tests"], "query_atom": "has_incident_note::Service A", "target_statement": "Service A has an incident note.", "gold_reason_type": "global_closed_world_underivable", "symbolic": {"positive_atoms": ["passed_unit_tests::Service A"], "negative_atoms": [], "rules": [], "query_atom": "has_incident_note::Service A", "closure_atoms": ["passed_unit_tests::Service A"]}, "source_reports": {"agent_a": {"positive": ["Service A passed unit tests."], "negative": []}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "closed_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_deployment_approval, has_incident_note, may_deploy, needs_sre_review, passed_unit_tests.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- has_deployment_approval: has deployment approval\n- has_incident_note: has an incident note\n- may_deploy: may deploy to production\n- needs_sre_review: needs SRE review\n- passed_unit_tests: passed unit tests\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- Service A passed unit tests.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: has_incident_note::Service A.\nTarget statement: Service A has an incident note.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "cloud_deployment_0246_open_missing_direct__lcwa__multi_agent", "source_id": "cloud_deployment_0246_open_missing_direct__lcwa", "base_id": "cloud_deployment_0246_open_missing_direct", "split": "train", "domain": "cloud_deployment", "family": "open_missing_direct", "subset": "core_contrastive", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_deployment_approval", "may_deploy"], "agent_a_complete_predicates": ["has_deployment_approval", "may_deploy"], "agent_b_open_predicates": ["has_incident_note", "needs_sre_review", "passed_unit_tests"], "facts_positive": ["Service A passed unit tests."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_unit_tests": "passed unit tests", "has_deployment_approval": "has deployment approval", "has_incident_note": "has an incident note", "may_deploy": "may deploy to production", "needs_sre_review": "needs SRE review"}, "vocabulary_predicates": ["has_deployment_approval", "has_incident_note", "may_deploy", "needs_sre_review", "passed_unit_tests"], "query_atom": "has_incident_note::Service A", "target_statement": "Service A has an incident note.", "gold_reason_type": "local_open_world_underivable", "symbolic": {"positive_atoms": ["passed_unit_tests::Service A"], "negative_atoms": [], "rules": [], "query_atom": "has_incident_note::Service A", "closure_atoms": ["passed_unit_tests::Service A"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Service A passed unit tests."], "negative": []}}, "gold_truth_value": "unknown", "gold_source_used": "agent_b", "gold_closure_handling": "open_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_deployment_approval, may_deploy.\nAgent B open/incomplete predicates: has_incident_note, needs_sre_review, passed_unit_tests.\n\nPredicate glossary:\n- has_deployment_approval: has deployment approval\n- has_incident_note: has an incident note\n- may_deploy: may deploy to production\n- needs_sre_review: needs SRE review\n- passed_unit_tests: passed unit tests\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Service A passed unit tests.\nNegative facts:\n- none\n\nTarget atom: has_incident_note::Service A.\nTarget statement: Service A has an incident note.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "cloud_deployment_0246_open_missing_direct__owa__multi_agent", "source_id": "cloud_deployment_0246_open_missing_direct__owa", "base_id": "cloud_deployment_0246_open_missing_direct", "split": "train", "domain": "cloud_deployment", "family": "open_missing_direct", "subset": "core_contrastive", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["has_deployment_approval", "has_incident_note", "may_deploy", "needs_sre_review", "passed_unit_tests"], "facts_positive": ["Service A passed unit tests."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_unit_tests": "passed unit tests", "has_deployment_approval": "has deployment approval", "has_incident_note": "has an incident note", "may_deploy": "may deploy to production", "needs_sre_review": "needs SRE review"}, "vocabulary_predicates": ["has_deployment_approval", "has_incident_note", "may_deploy", "needs_sre_review", "passed_unit_tests"], "query_atom": "has_incident_note::Service A", "target_statement": "Service A has an incident note.", "gold_reason_type": "open_world_underivable", "symbolic": {"positive_atoms": ["passed_unit_tests::Service A"], "negative_atoms": [], "rules": [], "query_atom": "has_incident_note::Service A", "closure_atoms": ["passed_unit_tests::Service A"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Service A passed unit tests."], "negative": []}}, "gold_truth_value": "unknown", "gold_source_used": "agent_b", "gold_closure_handling": "open_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: has_deployment_approval, has_incident_note, may_deploy, needs_sre_review, passed_unit_tests.\n\nPredicate glossary:\n- has_deployment_approval: has deployment approval\n- has_incident_note: has an incident note\n- may_deploy: may deploy to production\n- needs_sre_review: needs SRE review\n- passed_unit_tests: passed unit tests\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Service A passed unit tests.\nNegative facts:\n- none\n\nTarget atom: has_incident_note::Service A.\nTarget statement: Service A has an incident note.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "cloud_deployment_0248_closed_missing_with_open_distractor__cwa__multi_agent", "source_id": "cloud_deployment_0248_closed_missing_with_open_distractor__cwa", "base_id": "cloud_deployment_0248_closed_missing_with_open_distractor", "split": "train", "domain": "cloud_deployment", "family": "closed_missing_with_open_distractor", "subset": "core_contrastive", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["has_deployment_approval", "has_incident_note", "may_deploy", "needs_sre_review", "passed_unit_tests"], "agent_a_complete_predicates": ["has_deployment_approval", "has_incident_note", "may_deploy", "needs_sre_review", "passed_unit_tests"], "agent_b_open_predicates": [], "facts_positive": ["Service D has an incident note.", "Service C passed unit tests."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_unit_tests": "passed unit tests", "has_deployment_approval": "has deployment approval", "has_incident_note": "has an incident note", "may_deploy": "may deploy to production", "needs_sre_review": "needs SRE review"}, "vocabulary_predicates": ["has_deployment_approval", "has_incident_note", "may_deploy", "needs_sre_review", "passed_unit_tests"], "query_atom": "has_deployment_approval::Service C", "target_statement": "Service C has deployment approval.", "gold_reason_type": "global_closed_world_underivable", "symbolic": {"positive_atoms": ["has_incident_note::Service D", "passed_unit_tests::Service C"], "negative_atoms": [], "rules": [], "query_atom": "has_deployment_approval::Service C", "closure_atoms": ["has_incident_note::Service D", "passed_unit_tests::Service C"]}, "source_reports": {"agent_a": {"positive": ["Service D has an incident note.", "Service C passed unit tests."], "negative": []}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "closed_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_deployment_approval, has_incident_note, may_deploy, needs_sre_review, passed_unit_tests.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- has_deployment_approval: has deployment approval\n- has_incident_note: has an incident note\n- may_deploy: may deploy to production\n- needs_sre_review: needs SRE review\n- passed_unit_tests: passed unit tests\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- Service D has an incident note.\n- Service C passed unit tests.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: has_deployment_approval::Service C.\nTarget statement: Service C has deployment approval.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "cloud_deployment_0248_closed_missing_with_open_distractor__lcwa__multi_agent", "source_id": "cloud_deployment_0248_closed_missing_with_open_distractor__lcwa", "base_id": "cloud_deployment_0248_closed_missing_with_open_distractor", "split": "train", "domain": "cloud_deployment", "family": "closed_missing_with_open_distractor", "subset": "core_contrastive", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_deployment_approval"], "agent_a_complete_predicates": ["has_deployment_approval"], "agent_b_open_predicates": ["has_incident_note", "may_deploy", "needs_sre_review", "passed_unit_tests"], "facts_positive": ["Service D has an incident note.", "Service C passed unit tests."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_unit_tests": "passed unit tests", "has_deployment_approval": "has deployment approval", "has_incident_note": "has an incident note", "may_deploy": "may deploy to production", "needs_sre_review": "needs SRE review"}, "vocabulary_predicates": ["has_deployment_approval", "has_incident_note", "may_deploy", "needs_sre_review", "passed_unit_tests"], "query_atom": "has_deployment_approval::Service C", "target_statement": "Service C has deployment approval.", "gold_reason_type": "local_closed_world_underivable", "symbolic": {"positive_atoms": ["has_incident_note::Service D", "passed_unit_tests::Service C"], "negative_atoms": [], "rules": [], "query_atom": "has_deployment_approval::Service C", "closure_atoms": ["has_incident_note::Service D", "passed_unit_tests::Service C"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Service D has an incident note.", "Service C passed unit tests."], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "closed_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_deployment_approval.\nAgent B open/incomplete predicates: has_incident_note, may_deploy, needs_sre_review, passed_unit_tests.\n\nPredicate glossary:\n- has_deployment_approval: has deployment approval\n- has_incident_note: has an incident note\n- may_deploy: may deploy to production\n- needs_sre_review: needs SRE review\n- passed_unit_tests: passed unit tests\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Service D has an incident note.\n- Service C passed unit tests.\nNegative facts:\n- none\n\nTarget atom: has_deployment_approval::Service C.\nTarget statement: Service C has deployment approval.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "cloud_deployment_0248_closed_missing_with_open_distractor__owa__multi_agent", "source_id": "cloud_deployment_0248_closed_missing_with_open_distractor__owa", "base_id": "cloud_deployment_0248_closed_missing_with_open_distractor", "split": "train", "domain": "cloud_deployment", "family": "closed_missing_with_open_distractor", "subset": "core_contrastive", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["has_deployment_approval", "has_incident_note", "may_deploy", "needs_sre_review", "passed_unit_tests"], "facts_positive": ["Service D has an incident note.", "Service C passed unit tests."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_unit_tests": "passed unit tests", "has_deployment_approval": "has deployment approval", "has_incident_note": "has an incident note", "may_deploy": "may deploy to production", "needs_sre_review": "needs SRE review"}, "vocabulary_predicates": ["has_deployment_approval", "has_incident_note", "may_deploy", "needs_sre_review", "passed_unit_tests"], "query_atom": "has_deployment_approval::Service C", "target_statement": "Service C has deployment approval.", "gold_reason_type": "open_world_underivable", "symbolic": {"positive_atoms": ["has_incident_note::Service D", "passed_unit_tests::Service C"], "negative_atoms": [], "rules": [], "query_atom": "has_deployment_approval::Service C", "closure_atoms": ["has_incident_note::Service D", "passed_unit_tests::Service C"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Service D has an incident note.", "Service C passed unit tests."], "negative": []}}, "gold_truth_value": "unknown", "gold_source_used": "agent_b", "gold_closure_handling": "open_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: has_deployment_approval, has_incident_note, may_deploy, needs_sre_review, passed_unit_tests.\n\nPredicate glossary:\n- has_deployment_approval: has deployment approval\n- has_incident_note: has an incident note\n- may_deploy: may deploy to production\n- needs_sre_review: needs SRE review\n- passed_unit_tests: passed unit tests\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Service D has an incident note.\n- Service C passed unit tests.\nNegative facts:\n- none\n\nTarget atom: has_deployment_approval::Service C.\nTarget statement: Service C has deployment approval.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "cloud_deployment_0250_closed_missing_with_open_distractor__cwa__multi_agent", "source_id": "cloud_deployment_0250_closed_missing_with_open_distractor__cwa", "base_id": "cloud_deployment_0250_closed_missing_with_open_distractor", "split": "test", "domain": "cloud_deployment", "family": "closed_missing_with_open_distractor", "subset": "core_contrastive", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["has_deployment_approval", "has_incident_note", "may_deploy", "needs_sre_review", "passed_unit_tests"], "agent_a_complete_predicates": ["has_deployment_approval", "has_incident_note", "may_deploy", "needs_sre_review", "passed_unit_tests"], "agent_b_open_predicates": [], "facts_positive": ["Service F has an incident note.", "Service E passed unit tests."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_unit_tests": "passed unit tests", "has_deployment_approval": "has deployment approval", "has_incident_note": "has an incident note", "may_deploy": "may deploy to production", "needs_sre_review": "needs SRE review"}, "vocabulary_predicates": ["has_deployment_approval", "has_incident_note", "may_deploy", "needs_sre_review", "passed_unit_tests"], "query_atom": "has_deployment_approval::Service E", "target_statement": "Service E has deployment approval.", "gold_reason_type": "global_closed_world_underivable", "symbolic": {"positive_atoms": ["has_incident_note::Service F", "passed_unit_tests::Service E"], "negative_atoms": [], "rules": [], "query_atom": "has_deployment_approval::Service E", "closure_atoms": ["has_incident_note::Service F", "passed_unit_tests::Service E"]}, "source_reports": {"agent_a": {"positive": ["Service F has an incident note.", "Service E passed unit tests."], "negative": []}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "closed_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_deployment_approval, has_incident_note, may_deploy, needs_sre_review, passed_unit_tests.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- has_deployment_approval: has deployment approval\n- has_incident_note: has an incident note\n- may_deploy: may deploy to production\n- needs_sre_review: needs SRE review\n- passed_unit_tests: passed unit tests\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- Service F has an incident note.\n- Service E passed unit tests.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: has_deployment_approval::Service E.\nTarget statement: Service E has deployment approval.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "cloud_deployment_0250_closed_missing_with_open_distractor__lcwa__multi_agent", "source_id": "cloud_deployment_0250_closed_missing_with_open_distractor__lcwa", "base_id": "cloud_deployment_0250_closed_missing_with_open_distractor", "split": "test", "domain": "cloud_deployment", "family": "closed_missing_with_open_distractor", "subset": "core_contrastive", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_deployment_approval"], "agent_a_complete_predicates": ["has_deployment_approval"], "agent_b_open_predicates": ["has_incident_note", "may_deploy", "needs_sre_review", "passed_unit_tests"], "facts_positive": ["Service F has an incident note.", "Service E passed unit tests."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_unit_tests": "passed unit tests", "has_deployment_approval": "has deployment approval", "has_incident_note": "has an incident note", "may_deploy": "may deploy to production", "needs_sre_review": "needs SRE review"}, "vocabulary_predicates": ["has_deployment_approval", "has_incident_note", "may_deploy", "needs_sre_review", "passed_unit_tests"], "query_atom": "has_deployment_approval::Service E", "target_statement": "Service E has deployment approval.", "gold_reason_type": "local_closed_world_underivable", "symbolic": {"positive_atoms": ["has_incident_note::Service F", "passed_unit_tests::Service E"], "negative_atoms": [], "rules": [], "query_atom": "has_deployment_approval::Service E", "closure_atoms": ["has_incident_note::Service F", "passed_unit_tests::Service E"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Service F has an incident note.", "Service E passed unit tests."], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "closed_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_deployment_approval.\nAgent B open/incomplete predicates: has_incident_note, may_deploy, needs_sre_review, passed_unit_tests.\n\nPredicate glossary:\n- has_deployment_approval: has deployment approval\n- has_incident_note: has an incident note\n- may_deploy: may deploy to production\n- needs_sre_review: needs SRE review\n- passed_unit_tests: passed unit tests\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Service F has an incident note.\n- Service E passed unit tests.\nNegative facts:\n- none\n\nTarget atom: has_deployment_approval::Service E.\nTarget statement: Service E has deployment approval.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "cloud_deployment_0250_closed_missing_with_open_distractor__owa__multi_agent", "source_id": "cloud_deployment_0250_closed_missing_with_open_distractor__owa", "base_id": "cloud_deployment_0250_closed_missing_with_open_distractor", "split": "test", "domain": "cloud_deployment", "family": "closed_missing_with_open_distractor", "subset": "core_contrastive", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["has_deployment_approval", "has_incident_note", "may_deploy", "needs_sre_review", "passed_unit_tests"], "facts_positive": ["Service F has an incident note.", "Service E passed unit tests."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_unit_tests": "passed unit tests", "has_deployment_approval": "has deployment approval", "has_incident_note": "has an incident note", "may_deploy": "may deploy to production", "needs_sre_review": "needs SRE review"}, "vocabulary_predicates": ["has_deployment_approval", "has_incident_note", "may_deploy", "needs_sre_review", "passed_unit_tests"], "query_atom": "has_deployment_approval::Service E", "target_statement": "Service E has deployment approval.", "gold_reason_type": "open_world_underivable", "symbolic": {"positive_atoms": ["has_incident_note::Service F", "passed_unit_tests::Service E"], "negative_atoms": [], "rules": [], "query_atom": "has_deployment_approval::Service E", "closure_atoms": ["has_incident_note::Service F", "passed_unit_tests::Service E"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Service F has an incident note.", "Service E passed unit tests."], "negative": []}}, "gold_truth_value": "unknown", "gold_source_used": "agent_b", "gold_closure_handling": "open_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: has_deployment_approval, has_incident_note, may_deploy, needs_sre_review, passed_unit_tests.\n\nPredicate glossary:\n- has_deployment_approval: has deployment approval\n- has_incident_note: has an incident note\n- may_deploy: may deploy to production\n- needs_sre_review: needs SRE review\n- passed_unit_tests: passed unit tests\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Service F has an incident note.\n- Service E passed unit tests.\nNegative facts:\n- none\n\nTarget atom: has_deployment_approval::Service E.\nTarget statement: Service E has deployment approval.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "cloud_deployment_0251_closed_missing_with_open_distractor__cwa__multi_agent", "source_id": "cloud_deployment_0251_closed_missing_with_open_distractor__cwa", "base_id": "cloud_deployment_0251_closed_missing_with_open_distractor", "split": "train", "domain": "cloud_deployment", "family": "closed_missing_with_open_distractor", "subset": "core_contrastive", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["has_deployment_approval", "has_incident_note", "may_deploy", "needs_sre_review", "passed_unit_tests"], "agent_a_complete_predicates": ["has_deployment_approval", "has_incident_note", "may_deploy", "needs_sre_review", "passed_unit_tests"], "agent_b_open_predicates": [], "facts_positive": ["Service A has an incident note.", "Service F passed unit tests."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_unit_tests": "passed unit tests", "has_deployment_approval": "has deployment approval", "has_incident_note": "has an incident note", "may_deploy": "may deploy to production", "needs_sre_review": "needs SRE review"}, "vocabulary_predicates": ["has_deployment_approval", "has_incident_note", "may_deploy", "needs_sre_review", "passed_unit_tests"], "query_atom": "has_deployment_approval::Service F", "target_statement": "Service F has deployment approval.", "gold_reason_type": "global_closed_world_underivable", "symbolic": {"positive_atoms": ["has_incident_note::Service A", "passed_unit_tests::Service F"], "negative_atoms": [], "rules": [], "query_atom": "has_deployment_approval::Service F", "closure_atoms": ["has_incident_note::Service A", "passed_unit_tests::Service F"]}, "source_reports": {"agent_a": {"positive": ["Service A has an incident note.", "Service F passed unit tests."], "negative": []}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "closed_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_deployment_approval, has_incident_note, may_deploy, needs_sre_review, passed_unit_tests.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- has_deployment_approval: has deployment approval\n- has_incident_note: has an incident note\n- may_deploy: may deploy to production\n- needs_sre_review: needs SRE review\n- passed_unit_tests: passed unit tests\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- Service A has an incident note.\n- Service F passed unit tests.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: has_deployment_approval::Service F.\nTarget statement: Service F has deployment approval.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "cloud_deployment_0251_closed_missing_with_open_distractor__lcwa__multi_agent", "source_id": "cloud_deployment_0251_closed_missing_with_open_distractor__lcwa", "base_id": "cloud_deployment_0251_closed_missing_with_open_distractor", "split": "train", "domain": "cloud_deployment", "family": "closed_missing_with_open_distractor", "subset": "core_contrastive", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_deployment_approval"], "agent_a_complete_predicates": ["has_deployment_approval"], "agent_b_open_predicates": ["has_incident_note", "may_deploy", "needs_sre_review", "passed_unit_tests"], "facts_positive": ["Service A has an incident note.", "Service F passed unit tests."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_unit_tests": "passed unit tests", "has_deployment_approval": "has deployment approval", "has_incident_note": "has an incident note", "may_deploy": "may deploy to production", "needs_sre_review": "needs SRE review"}, "vocabulary_predicates": ["has_deployment_approval", "has_incident_note", "may_deploy", "needs_sre_review", "passed_unit_tests"], "query_atom": "has_deployment_approval::Service F", "target_statement": "Service F has deployment approval.", "gold_reason_type": "local_closed_world_underivable", "symbolic": {"positive_atoms": ["has_incident_note::Service A", "passed_unit_tests::Service F"], "negative_atoms": [], "rules": [], "query_atom": "has_deployment_approval::Service F", "closure_atoms": ["has_incident_note::Service A", "passed_unit_tests::Service F"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Service A has an incident note.", "Service F passed unit tests."], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "closed_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_deployment_approval.\nAgent B open/incomplete predicates: has_incident_note, may_deploy, needs_sre_review, passed_unit_tests.\n\nPredicate glossary:\n- has_deployment_approval: has deployment approval\n- has_incident_note: has an incident note\n- may_deploy: may deploy to production\n- needs_sre_review: needs SRE review\n- passed_unit_tests: passed unit tests\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Service A has an incident note.\n- Service F passed unit tests.\nNegative facts:\n- none\n\nTarget atom: has_deployment_approval::Service F.\nTarget statement: Service F has deployment approval.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "cloud_deployment_0251_closed_missing_with_open_distractor__owa__multi_agent", "source_id": "cloud_deployment_0251_closed_missing_with_open_distractor__owa", "base_id": "cloud_deployment_0251_closed_missing_with_open_distractor", "split": "train", "domain": "cloud_deployment", "family": "closed_missing_with_open_distractor", "subset": "core_contrastive", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["has_deployment_approval", "has_incident_note", "may_deploy", "needs_sre_review", "passed_unit_tests"], "facts_positive": ["Service A has an incident note.", "Service F passed unit tests."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_unit_tests": "passed unit tests", "has_deployment_approval": "has deployment approval", "has_incident_note": "has an incident note", "may_deploy": "may deploy to production", "needs_sre_review": "needs SRE review"}, "vocabulary_predicates": ["has_deployment_approval", "has_incident_note", "may_deploy", "needs_sre_review", "passed_unit_tests"], "query_atom": "has_deployment_approval::Service F", "target_statement": "Service F has deployment approval.", "gold_reason_type": "open_world_underivable", "symbolic": {"positive_atoms": ["has_incident_note::Service A", "passed_unit_tests::Service F"], "negative_atoms": [], "rules": [], "query_atom": "has_deployment_approval::Service F", "closure_atoms": ["has_incident_note::Service A", "passed_unit_tests::Service F"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Service A has an incident note.", "Service F passed unit tests."], "negative": []}}, "gold_truth_value": "unknown", "gold_source_used": "agent_b", "gold_closure_handling": "open_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: has_deployment_approval, has_incident_note, may_deploy, needs_sre_review, passed_unit_tests.\n\nPredicate glossary:\n- has_deployment_approval: has deployment approval\n- has_incident_note: has an incident note\n- may_deploy: may deploy to production\n- needs_sre_review: needs SRE review\n- passed_unit_tests: passed unit tests\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Service A has an incident note.\n- Service F passed unit tests.\nNegative facts:\n- none\n\nTarget atom: has_deployment_approval::Service F.\nTarget statement: Service F has deployment approval.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "cloud_deployment_0252_open_missing_with_closed_distractor__cwa__multi_agent", "source_id": "cloud_deployment_0252_open_missing_with_closed_distractor__cwa", "base_id": "cloud_deployment_0252_open_missing_with_closed_distractor", "split": "test", "domain": "cloud_deployment", "family": "open_missing_with_closed_distractor", "subset": "core_contrastive", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["has_deployment_approval", "has_incident_note", "may_deploy", "needs_sre_review", "passed_unit_tests"], "agent_a_complete_predicates": ["has_deployment_approval", "has_incident_note", "may_deploy", "needs_sre_review", "passed_unit_tests"], "agent_b_open_predicates": [], "facts_positive": ["Service B has deployment approval.", "Service B passed unit tests."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_unit_tests": "passed unit tests", "has_deployment_approval": "has deployment approval", "has_incident_note": "has an incident note", "may_deploy": "may deploy to production", "needs_sre_review": "needs SRE review"}, "vocabulary_predicates": ["has_deployment_approval", "has_incident_note", "may_deploy", "needs_sre_review", "passed_unit_tests"], "query_atom": "has_incident_note::Service A", "target_statement": "Service A has an incident note.", "gold_reason_type": "global_closed_world_underivable", "symbolic": {"positive_atoms": ["has_deployment_approval::Service B", "passed_unit_tests::Service B"], "negative_atoms": [], "rules": [], "query_atom": "has_incident_note::Service A", "closure_atoms": ["has_deployment_approval::Service B", "passed_unit_tests::Service B"]}, "source_reports": {"agent_a": {"positive": ["Service B has deployment approval.", "Service B passed unit tests."], "negative": []}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "closed_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_deployment_approval, has_incident_note, may_deploy, needs_sre_review, passed_unit_tests.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- has_deployment_approval: has deployment approval\n- has_incident_note: has an incident note\n- may_deploy: may deploy to production\n- needs_sre_review: needs SRE review\n- passed_unit_tests: passed unit tests\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- Service B has deployment approval.\n- Service B passed unit tests.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: has_incident_note::Service A.\nTarget statement: Service A has an incident note.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "cloud_deployment_0252_open_missing_with_closed_distractor__lcwa__multi_agent", "source_id": "cloud_deployment_0252_open_missing_with_closed_distractor__lcwa", "base_id": "cloud_deployment_0252_open_missing_with_closed_distractor", "split": "test", "domain": "cloud_deployment", "family": "open_missing_with_closed_distractor", "subset": "core_contrastive", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_deployment_approval", "may_deploy"], "agent_a_complete_predicates": ["has_deployment_approval", "may_deploy"], "agent_b_open_predicates": ["has_incident_note", "needs_sre_review", "passed_unit_tests"], "facts_positive": ["Service B has deployment approval.", "Service B passed unit tests."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_unit_tests": "passed unit tests", "has_deployment_approval": "has deployment approval", "has_incident_note": "has an incident note", "may_deploy": "may deploy to production", "needs_sre_review": "needs SRE review"}, "vocabulary_predicates": ["has_deployment_approval", "has_incident_note", "may_deploy", "needs_sre_review", "passed_unit_tests"], "query_atom": "has_incident_note::Service A", "target_statement": "Service A has an incident note.", "gold_reason_type": "local_open_world_underivable", "symbolic": {"positive_atoms": ["has_deployment_approval::Service B", "passed_unit_tests::Service B"], "negative_atoms": [], "rules": [], "query_atom": "has_incident_note::Service A", "closure_atoms": ["has_deployment_approval::Service B", "passed_unit_tests::Service B"]}, "source_reports": {"agent_a": {"positive": ["Service B has deployment approval."], "negative": []}, "agent_b": {"positive": ["Service B passed unit tests."], "negative": []}}, "gold_truth_value": "unknown", "gold_source_used": "agent_b", "gold_closure_handling": "open_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_deployment_approval, may_deploy.\nAgent B open/incomplete predicates: has_incident_note, needs_sre_review, passed_unit_tests.\n\nPredicate glossary:\n- has_deployment_approval: has deployment approval\n- has_incident_note: has an incident note\n- may_deploy: may deploy to production\n- needs_sre_review: needs SRE review\n- passed_unit_tests: passed unit tests\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- Service B has deployment approval.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Service B passed unit tests.\nNegative facts:\n- none\n\nTarget atom: has_incident_note::Service A.\nTarget statement: Service A has an incident note.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "cloud_deployment_0252_open_missing_with_closed_distractor__owa__multi_agent", "source_id": "cloud_deployment_0252_open_missing_with_closed_distractor__owa", "base_id": "cloud_deployment_0252_open_missing_with_closed_distractor", "split": "test", "domain": "cloud_deployment", "family": "open_missing_with_closed_distractor", "subset": "core_contrastive", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["has_deployment_approval", "has_incident_note", "may_deploy", "needs_sre_review", "passed_unit_tests"], "facts_positive": ["Service B has deployment approval.", "Service B passed unit tests."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_unit_tests": "passed unit tests", "has_deployment_approval": "has deployment approval", "has_incident_note": "has an incident note", "may_deploy": "may deploy to production", "needs_sre_review": "needs SRE review"}, "vocabulary_predicates": ["has_deployment_approval", "has_incident_note", "may_deploy", "needs_sre_review", "passed_unit_tests"], "query_atom": "has_incident_note::Service A", "target_statement": "Service A has an incident note.", "gold_reason_type": "open_world_underivable", "symbolic": {"positive_atoms": ["has_deployment_approval::Service B", "passed_unit_tests::Service B"], "negative_atoms": [], "rules": [], "query_atom": "has_incident_note::Service A", "closure_atoms": ["has_deployment_approval::Service B", "passed_unit_tests::Service B"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Service B has deployment approval.", "Service B passed unit tests."], "negative": []}}, "gold_truth_value": "unknown", "gold_source_used": "agent_b", "gold_closure_handling": "open_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: has_deployment_approval, has_incident_note, may_deploy, needs_sre_review, passed_unit_tests.\n\nPredicate glossary:\n- has_deployment_approval: has deployment approval\n- has_incident_note: has an incident note\n- may_deploy: may deploy to production\n- needs_sre_review: needs SRE review\n- passed_unit_tests: passed unit tests\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Service B has deployment approval.\n- Service B passed unit tests.\nNegative facts:\n- none\n\nTarget atom: has_incident_note::Service A.\nTarget statement: Service A has an incident note.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "cloud_deployment_0253_open_missing_with_closed_distractor__cwa__multi_agent", "source_id": "cloud_deployment_0253_open_missing_with_closed_distractor__cwa", "base_id": "cloud_deployment_0253_open_missing_with_closed_distractor", "split": "train", "domain": "cloud_deployment", "family": "open_missing_with_closed_distractor", "subset": "core_contrastive", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["has_deployment_approval", "has_incident_note", "may_deploy", "needs_sre_review", "passed_unit_tests"], "agent_a_complete_predicates": ["has_deployment_approval", "has_incident_note", "may_deploy", "needs_sre_review", "passed_unit_tests"], "agent_b_open_predicates": [], "facts_positive": ["Service C has deployment approval.", "Service C passed unit tests."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_unit_tests": "passed unit tests", "has_deployment_approval": "has deployment approval", "has_incident_note": "has an incident note", "may_deploy": "may deploy to production", "needs_sre_review": "needs SRE review"}, "vocabulary_predicates": ["has_deployment_approval", "has_incident_note", "may_deploy", "needs_sre_review", "passed_unit_tests"], "query_atom": "has_incident_note::Service B", "target_statement": "Service B has an incident note.", "gold_reason_type": "global_closed_world_underivable", "symbolic": {"positive_atoms": ["has_deployment_approval::Service C", "passed_unit_tests::Service C"], "negative_atoms": [], "rules": [], "query_atom": "has_incident_note::Service B", "closure_atoms": ["has_deployment_approval::Service C", "passed_unit_tests::Service C"]}, "source_reports": {"agent_a": {"positive": ["Service C has deployment approval.", "Service C passed unit tests."], "negative": []}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "closed_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_deployment_approval, has_incident_note, may_deploy, needs_sre_review, passed_unit_tests.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- has_deployment_approval: has deployment approval\n- has_incident_note: has an incident note\n- may_deploy: may deploy to production\n- needs_sre_review: needs SRE review\n- passed_unit_tests: passed unit tests\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- Service C has deployment approval.\n- Service C passed unit tests.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: has_incident_note::Service B.\nTarget statement: Service B has an incident note.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "cloud_deployment_0253_open_missing_with_closed_distractor__lcwa__multi_agent", "source_id": "cloud_deployment_0253_open_missing_with_closed_distractor__lcwa", "base_id": "cloud_deployment_0253_open_missing_with_closed_distractor", "split": "train", "domain": "cloud_deployment", "family": "open_missing_with_closed_distractor", "subset": "core_contrastive", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_deployment_approval", "may_deploy"], "agent_a_complete_predicates": ["has_deployment_approval", "may_deploy"], "agent_b_open_predicates": ["has_incident_note", "needs_sre_review", "passed_unit_tests"], "facts_positive": ["Service C has deployment approval.", "Service C passed unit tests."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_unit_tests": "passed unit tests", "has_deployment_approval": "has deployment approval", "has_incident_note": "has an incident note", "may_deploy": "may deploy to production", "needs_sre_review": "needs SRE review"}, "vocabulary_predicates": ["has_deployment_approval", "has_incident_note", "may_deploy", "needs_sre_review", "passed_unit_tests"], "query_atom": "has_incident_note::Service B", "target_statement": "Service B has an incident note.", "gold_reason_type": "local_open_world_underivable", "symbolic": {"positive_atoms": ["has_deployment_approval::Service C", "passed_unit_tests::Service C"], "negative_atoms": [], "rules": [], "query_atom": "has_incident_note::Service B", "closure_atoms": ["has_deployment_approval::Service C", "passed_unit_tests::Service C"]}, "source_reports": {"agent_a": {"positive": ["Service C has deployment approval."], "negative": []}, "agent_b": {"positive": ["Service C passed unit tests."], "negative": []}}, "gold_truth_value": "unknown", "gold_source_used": "agent_b", "gold_closure_handling": "open_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_deployment_approval, may_deploy.\nAgent B open/incomplete predicates: has_incident_note, needs_sre_review, passed_unit_tests.\n\nPredicate glossary:\n- has_deployment_approval: has deployment approval\n- has_incident_note: has an incident note\n- may_deploy: may deploy to production\n- needs_sre_review: needs SRE review\n- passed_unit_tests: passed unit tests\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- Service C has deployment approval.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Service C passed unit tests.\nNegative facts:\n- none\n\nTarget atom: has_incident_note::Service B.\nTarget statement: Service B has an incident note.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "cloud_deployment_0253_open_missing_with_closed_distractor__owa__multi_agent", "source_id": "cloud_deployment_0253_open_missing_with_closed_distractor__owa", "base_id": "cloud_deployment_0253_open_missing_with_closed_distractor", "split": "train", "domain": "cloud_deployment", "family": "open_missing_with_closed_distractor", "subset": "core_contrastive", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["has_deployment_approval", "has_incident_note", "may_deploy", "needs_sre_review", "passed_unit_tests"], "facts_positive": ["Service C has deployment approval.", "Service C passed unit tests."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_unit_tests": "passed unit tests", "has_deployment_approval": "has deployment approval", "has_incident_note": "has an incident note", "may_deploy": "may deploy to production", "needs_sre_review": "needs SRE review"}, "vocabulary_predicates": ["has_deployment_approval", "has_incident_note", "may_deploy", "needs_sre_review", "passed_unit_tests"], "query_atom": "has_incident_note::Service B", "target_statement": "Service B has an incident note.", "gold_reason_type": "open_world_underivable", "symbolic": {"positive_atoms": ["has_deployment_approval::Service C", "passed_unit_tests::Service C"], "negative_atoms": [], "rules": [], "query_atom": "has_incident_note::Service B", "closure_atoms": ["has_deployment_approval::Service C", "passed_unit_tests::Service C"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Service C has deployment approval.", "Service C passed unit tests."], "negative": []}}, "gold_truth_value": "unknown", "gold_source_used": "agent_b", "gold_closure_handling": "open_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: has_deployment_approval, has_incident_note, may_deploy, needs_sre_review, passed_unit_tests.\n\nPredicate glossary:\n- has_deployment_approval: has deployment approval\n- has_incident_note: has an incident note\n- may_deploy: may deploy to production\n- needs_sre_review: needs SRE review\n- passed_unit_tests: passed unit tests\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Service C has deployment approval.\n- Service C passed unit tests.\nNegative facts:\n- none\n\nTarget atom: has_incident_note::Service B.\nTarget statement: Service B has an incident note.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "cloud_deployment_0258_closed_derived_missing_antecedent__cwa__multi_agent", "source_id": "cloud_deployment_0258_closed_derived_missing_antecedent__cwa", "base_id": "cloud_deployment_0258_closed_derived_missing_antecedent", "split": "train", "domain": "cloud_deployment", "family": "closed_derived_missing_antecedent", "subset": "core_contrastive", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["has_deployment_approval", "has_incident_note", "may_deploy", "needs_sre_review", "passed_unit_tests"], "agent_a_complete_predicates": ["has_deployment_approval", "has_incident_note", "may_deploy", "needs_sre_review", "passed_unit_tests"], "agent_b_open_predicates": [], "facts_positive": ["Service A passed unit tests."], "facts_negative": [], "rules_natural": ["If a service passed unit tests and has deployment approval, then that service may deploy to production."], "predicate_glossary": {"passed_unit_tests": "passed unit tests", "has_deployment_approval": "has deployment approval", "has_incident_note": "has an incident note", "may_deploy": "may deploy to production", "needs_sre_review": "needs SRE review"}, "vocabulary_predicates": ["has_deployment_approval", "has_incident_note", "may_deploy", "needs_sre_review", "passed_unit_tests"], "query_atom": "may_deploy::Service A", "target_statement": "Service A may deploy to production.", "gold_reason_type": "global_closed_world_underivable", "symbolic": {"positive_atoms": ["passed_unit_tests::Service A"], "negative_atoms": [], "rules": [{"antecedents": ["passed_unit_tests::Service A", "has_deployment_approval::Service A"], "conclusion": "may_deploy::Service A", "text": "If a service passed unit tests and has deployment approval, then that service may deploy to production."}], "query_atom": "may_deploy::Service A", "closure_atoms": ["passed_unit_tests::Service A"]}, "source_reports": {"agent_a": {"positive": ["Service A passed unit tests."], "negative": []}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "closed_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_deployment_approval, has_incident_note, may_deploy, needs_sre_review, passed_unit_tests.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- has_deployment_approval: has deployment approval\n- has_incident_note: has an incident note\n- may_deploy: may deploy to production\n- needs_sre_review: needs SRE review\n- passed_unit_tests: passed unit tests\n\nCoordinator policy rules:\n- If a service passed unit tests and has deployment approval, then that service may deploy to production.\n\nAgent A report:\nPositive facts:\n- Service A passed unit tests.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: may_deploy::Service A.\nTarget statement: Service A may deploy to production.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "cloud_deployment_0258_closed_derived_missing_antecedent__lcwa__multi_agent", "source_id": "cloud_deployment_0258_closed_derived_missing_antecedent__lcwa", "base_id": "cloud_deployment_0258_closed_derived_missing_antecedent", "split": "train", "domain": "cloud_deployment", "family": "closed_derived_missing_antecedent", "subset": "core_contrastive", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_deployment_approval", "may_deploy"], "agent_a_complete_predicates": ["has_deployment_approval", "may_deploy"], "agent_b_open_predicates": ["has_incident_note", "needs_sre_review", "passed_unit_tests"], "facts_positive": ["Service A passed unit tests."], "facts_negative": [], "rules_natural": ["If a service passed unit tests and has deployment approval, then that service may deploy to production."], "predicate_glossary": {"passed_unit_tests": "passed unit tests", "has_deployment_approval": "has deployment approval", "has_incident_note": "has an incident note", "may_deploy": "may deploy to production", "needs_sre_review": "needs SRE review"}, "vocabulary_predicates": ["has_deployment_approval", "has_incident_note", "may_deploy", "needs_sre_review", "passed_unit_tests"], "query_atom": "may_deploy::Service A", "target_statement": "Service A may deploy to production.", "gold_reason_type": "local_closed_world_underivable", "symbolic": {"positive_atoms": ["passed_unit_tests::Service A"], "negative_atoms": [], "rules": [{"antecedents": ["passed_unit_tests::Service A", "has_deployment_approval::Service A"], "conclusion": "may_deploy::Service A", "text": "If a service passed unit tests and has deployment approval, then that service may deploy to production."}], "query_atom": "may_deploy::Service A", "closure_atoms": ["passed_unit_tests::Service A"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Service A passed unit tests."], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "closed_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_deployment_approval, may_deploy.\nAgent B open/incomplete predicates: has_incident_note, needs_sre_review, passed_unit_tests.\n\nPredicate glossary:\n- has_deployment_approval: has deployment approval\n- has_incident_note: has an incident note\n- may_deploy: may deploy to production\n- needs_sre_review: needs SRE review\n- passed_unit_tests: passed unit tests\n\nCoordinator policy rules:\n- If a service passed unit tests and has deployment approval, then that service may deploy to production.\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Service A passed unit tests.\nNegative facts:\n- none\n\nTarget atom: may_deploy::Service A.\nTarget statement: Service A may deploy to production.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "cloud_deployment_0258_closed_derived_missing_antecedent__owa__multi_agent", "source_id": "cloud_deployment_0258_closed_derived_missing_antecedent__owa", "base_id": "cloud_deployment_0258_closed_derived_missing_antecedent", "split": "train", "domain": "cloud_deployment", "family": "closed_derived_missing_antecedent", "subset": "core_contrastive", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["has_deployment_approval", "has_incident_note", "may_deploy", "needs_sre_review", "passed_unit_tests"], "facts_positive": ["Service A passed unit tests."], "facts_negative": [], "rules_natural": ["If a service passed unit tests and has deployment approval, then that service may deploy to production."], "predicate_glossary": {"passed_unit_tests": "passed unit tests", "has_deployment_approval": "has deployment approval", "has_incident_note": "has an incident note", "may_deploy": "may deploy to production", "needs_sre_review": "needs SRE review"}, "vocabulary_predicates": ["has_deployment_approval", "has_incident_note", "may_deploy", "needs_sre_review", "passed_unit_tests"], "query_atom": "may_deploy::Service A", "target_statement": "Service A may deploy to production.", "gold_reason_type": "open_world_underivable", "symbolic": {"positive_atoms": ["passed_unit_tests::Service A"], "negative_atoms": [], "rules": [{"antecedents": ["passed_unit_tests::Service A", "has_deployment_approval::Service A"], "conclusion": "may_deploy::Service A", "text": "If a service passed unit tests and has deployment approval, then that service may deploy to production."}], "query_atom": "may_deploy::Service A", "closure_atoms": ["passed_unit_tests::Service A"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Service A passed unit tests."], "negative": []}}, "gold_truth_value": "unknown", "gold_source_used": "agent_b", "gold_closure_handling": "open_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: has_deployment_approval, has_incident_note, may_deploy, needs_sre_review, passed_unit_tests.\n\nPredicate glossary:\n- has_deployment_approval: has deployment approval\n- has_incident_note: has an incident note\n- may_deploy: may deploy to production\n- needs_sre_review: needs SRE review\n- passed_unit_tests: passed unit tests\n\nCoordinator policy rules:\n- If a service passed unit tests and has deployment approval, then that service may deploy to production.\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Service A passed unit tests.\nNegative facts:\n- none\n\nTarget atom: may_deploy::Service A.\nTarget statement: Service A may deploy to production.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "cloud_deployment_0263_open_derived_missing_antecedent__cwa__multi_agent", "source_id": "cloud_deployment_0263_open_derived_missing_antecedent__cwa", "base_id": "cloud_deployment_0263_open_derived_missing_antecedent", "split": "train", "domain": "cloud_deployment", "family": "open_derived_missing_antecedent", "subset": "core_contrastive", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["has_deployment_approval", "has_incident_note", "may_deploy", "needs_sre_review", "passed_unit_tests"], "agent_a_complete_predicates": ["has_deployment_approval", "has_incident_note", "may_deploy", "needs_sre_review", "passed_unit_tests"], "agent_b_open_predicates": [], "facts_positive": ["Service F passed unit tests."], "facts_negative": [], "rules_natural": ["If a service has an incident note, then that service needs SRE review."], "predicate_glossary": {"passed_unit_tests": "passed unit tests", "has_deployment_approval": "has deployment approval", "has_incident_note": "has an incident note", "may_deploy": "may deploy to production", "needs_sre_review": "needs SRE review"}, "vocabulary_predicates": ["has_deployment_approval", "has_incident_note", "may_deploy", "needs_sre_review", "passed_unit_tests"], "query_atom": "needs_sre_review::Service F", "target_statement": "Service F needs SRE review.", "gold_reason_type": "global_closed_world_underivable", "symbolic": {"positive_atoms": ["passed_unit_tests::Service F"], "negative_atoms": [], "rules": [{"antecedents": ["has_incident_note::Service F"], "conclusion": "needs_sre_review::Service F", "text": "If a service has an incident note, then that service needs SRE review."}], "query_atom": "needs_sre_review::Service F", "closure_atoms": ["passed_unit_tests::Service F"]}, "source_reports": {"agent_a": {"positive": ["Service F passed unit tests."], "negative": []}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "closed_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_deployment_approval, has_incident_note, may_deploy, needs_sre_review, passed_unit_tests.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- has_deployment_approval: has deployment approval\n- has_incident_note: has an incident note\n- may_deploy: may deploy to production\n- needs_sre_review: needs SRE review\n- passed_unit_tests: passed unit tests\n\nCoordinator policy rules:\n- If a service has an incident note, then that service needs SRE review.\n\nAgent A report:\nPositive facts:\n- Service F passed unit tests.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: needs_sre_review::Service F.\nTarget statement: Service F needs SRE review.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "cloud_deployment_0263_open_derived_missing_antecedent__lcwa__multi_agent", "source_id": "cloud_deployment_0263_open_derived_missing_antecedent__lcwa", "base_id": "cloud_deployment_0263_open_derived_missing_antecedent", "split": "train", "domain": "cloud_deployment", "family": "open_derived_missing_antecedent", "subset": "core_contrastive", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_deployment_approval", "may_deploy"], "agent_a_complete_predicates": ["has_deployment_approval", "may_deploy"], "agent_b_open_predicates": ["has_incident_note", "needs_sre_review", "passed_unit_tests"], "facts_positive": ["Service F passed unit tests."], "facts_negative": [], "rules_natural": ["If a service has an incident note, then that service needs SRE review."], "predicate_glossary": {"passed_unit_tests": "passed unit tests", "has_deployment_approval": "has deployment approval", "has_incident_note": "has an incident note", "may_deploy": "may deploy to production", "needs_sre_review": "needs SRE review"}, "vocabulary_predicates": ["has_deployment_approval", "has_incident_note", "may_deploy", "needs_sre_review", "passed_unit_tests"], "query_atom": "needs_sre_review::Service F", "target_statement": "Service F needs SRE review.", "gold_reason_type": "local_open_world_underivable", "symbolic": {"positive_atoms": ["passed_unit_tests::Service F"], "negative_atoms": [], "rules": [{"antecedents": ["has_incident_note::Service F"], "conclusion": "needs_sre_review::Service F", "text": "If a service has an incident note, then that service needs SRE review."}], "query_atom": "needs_sre_review::Service F", "closure_atoms": ["passed_unit_tests::Service F"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Service F passed unit tests."], "negative": []}}, "gold_truth_value": "unknown", "gold_source_used": "agent_b", "gold_closure_handling": "open_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_deployment_approval, may_deploy.\nAgent B open/incomplete predicates: has_incident_note, needs_sre_review, passed_unit_tests.\n\nPredicate glossary:\n- has_deployment_approval: has deployment approval\n- has_incident_note: has an incident note\n- may_deploy: may deploy to production\n- needs_sre_review: needs SRE review\n- passed_unit_tests: passed unit tests\n\nCoordinator policy rules:\n- If a service has an incident note, then that service needs SRE review.\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Service F passed unit tests.\nNegative facts:\n- none\n\nTarget atom: needs_sre_review::Service F.\nTarget statement: Service F needs SRE review.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "cloud_deployment_0263_open_derived_missing_antecedent__owa__multi_agent", "source_id": "cloud_deployment_0263_open_derived_missing_antecedent__owa", "base_id": "cloud_deployment_0263_open_derived_missing_antecedent", "split": "train", "domain": "cloud_deployment", "family": "open_derived_missing_antecedent", "subset": "core_contrastive", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["has_deployment_approval", "has_incident_note", "may_deploy", "needs_sre_review", "passed_unit_tests"], "facts_positive": ["Service F passed unit tests."], "facts_negative": [], "rules_natural": ["If a service has an incident note, then that service needs SRE review."], "predicate_glossary": {"passed_unit_tests": "passed unit tests", "has_deployment_approval": "has deployment approval", "has_incident_note": "has an incident note", "may_deploy": "may deploy to production", "needs_sre_review": "needs SRE review"}, "vocabulary_predicates": ["has_deployment_approval", "has_incident_note", "may_deploy", "needs_sre_review", "passed_unit_tests"], "query_atom": "needs_sre_review::Service F", "target_statement": "Service F needs SRE review.", "gold_reason_type": "open_world_underivable", "symbolic": {"positive_atoms": ["passed_unit_tests::Service F"], "negative_atoms": [], "rules": [{"antecedents": ["has_incident_note::Service F"], "conclusion": "needs_sre_review::Service F", "text": "If a service has an incident note, then that service needs SRE review."}], "query_atom": "needs_sre_review::Service F", "closure_atoms": ["passed_unit_tests::Service F"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Service F passed unit tests."], "negative": []}}, "gold_truth_value": "unknown", "gold_source_used": "agent_b", "gold_closure_handling": "open_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: has_deployment_approval, has_incident_note, may_deploy, needs_sre_review, passed_unit_tests.\n\nPredicate glossary:\n- has_deployment_approval: has deployment approval\n- has_incident_note: has an incident note\n- may_deploy: may deploy to production\n- needs_sre_review: needs SRE review\n- passed_unit_tests: passed unit tests\n\nCoordinator policy rules:\n- If a service has an incident note, then that service needs SRE review.\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Service F passed unit tests.\nNegative facts:\n- none\n\nTarget atom: needs_sre_review::Service F.\nTarget statement: Service F needs SRE review.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "cloud_deployment_0266_entailed_closed_conclusion__cwa__multi_agent", "source_id": "cloud_deployment_0266_entailed_closed_conclusion__cwa", "base_id": "cloud_deployment_0266_entailed_closed_conclusion", "split": "train", "domain": "cloud_deployment", "family": "entailed_closed_conclusion", "subset": "control_entailed", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["has_deployment_approval", "has_incident_note", "may_deploy", "needs_sre_review", "passed_unit_tests"], "agent_a_complete_predicates": ["has_deployment_approval", "has_incident_note", "may_deploy", "needs_sre_review", "passed_unit_tests"], "agent_b_open_predicates": [], "facts_positive": ["Service C has deployment approval.", "Service C passed unit tests."], "facts_negative": [], "rules_natural": ["If a service passed unit tests and has deployment approval, then that service may deploy to production."], "predicate_glossary": {"passed_unit_tests": "passed unit tests", "has_deployment_approval": "has deployment approval", "has_incident_note": "has an incident note", "may_deploy": "may deploy to production", "needs_sre_review": "needs SRE review"}, "vocabulary_predicates": ["has_deployment_approval", "has_incident_note", "may_deploy", "needs_sre_review", "passed_unit_tests"], "query_atom": "may_deploy::Service C", "target_statement": "Service C may deploy to production.", "gold_reason_type": "entailed_by_fact_or_rule", "symbolic": {"positive_atoms": ["has_deployment_approval::Service C", "passed_unit_tests::Service C"], "negative_atoms": [], "rules": [{"antecedents": ["passed_unit_tests::Service C", "has_deployment_approval::Service C"], "conclusion": "may_deploy::Service C", "text": "If a service passed unit tests and has deployment approval, then that service may deploy to production."}], "query_atom": "may_deploy::Service C", "closure_atoms": ["has_deployment_approval::Service C", "may_deploy::Service C", "passed_unit_tests::Service C"]}, "source_reports": {"agent_a": {"positive": ["Service C has deployment approval.", "Service C passed unit tests."], "negative": []}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "true", "gold_source_used": "agent_a", "gold_closure_handling": "entailed_or_explicit", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_deployment_approval, has_incident_note, may_deploy, needs_sre_review, passed_unit_tests.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- has_deployment_approval: has deployment approval\n- has_incident_note: has an incident note\n- may_deploy: may deploy to production\n- needs_sre_review: needs SRE review\n- passed_unit_tests: passed unit tests\n\nCoordinator policy rules:\n- If a service passed unit tests and has deployment approval, then that service may deploy to production.\n\nAgent A report:\nPositive facts:\n- Service C has deployment approval.\n- Service C passed unit tests.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: may_deploy::Service C.\nTarget statement: Service C may deploy to production.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "cloud_deployment_0266_entailed_closed_conclusion__lcwa__multi_agent", "source_id": "cloud_deployment_0266_entailed_closed_conclusion__lcwa", "base_id": "cloud_deployment_0266_entailed_closed_conclusion", "split": "train", "domain": "cloud_deployment", "family": "entailed_closed_conclusion", "subset": "control_entailed", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_deployment_approval", "may_deploy"], "agent_a_complete_predicates": ["has_deployment_approval", "may_deploy"], "agent_b_open_predicates": ["has_incident_note", "needs_sre_review", "passed_unit_tests"], "facts_positive": ["Service C has deployment approval.", "Service C passed unit tests."], "facts_negative": [], "rules_natural": ["If a service passed unit tests and has deployment approval, then that service may deploy to production."], "predicate_glossary": {"passed_unit_tests": "passed unit tests", "has_deployment_approval": "has deployment approval", "has_incident_note": "has an incident note", "may_deploy": "may deploy to production", "needs_sre_review": "needs SRE review"}, "vocabulary_predicates": ["has_deployment_approval", "has_incident_note", "may_deploy", "needs_sre_review", "passed_unit_tests"], "query_atom": "may_deploy::Service C", "target_statement": "Service C may deploy to production.", "gold_reason_type": "entailed_by_fact_or_rule", "symbolic": {"positive_atoms": ["has_deployment_approval::Service C", "passed_unit_tests::Service C"], "negative_atoms": [], "rules": [{"antecedents": ["passed_unit_tests::Service C", "has_deployment_approval::Service C"], "conclusion": "may_deploy::Service C", "text": "If a service passed unit tests and has deployment approval, then that service may deploy to production."}], "query_atom": "may_deploy::Service C", "closure_atoms": ["has_deployment_approval::Service C", "may_deploy::Service C", "passed_unit_tests::Service C"]}, "source_reports": {"agent_a": {"positive": ["Service C has deployment approval."], "negative": []}, "agent_b": {"positive": ["Service C passed unit tests."], "negative": []}}, "gold_truth_value": "true", "gold_source_used": "both", "gold_closure_handling": "entailed_or_explicit", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_deployment_approval, may_deploy.\nAgent B open/incomplete predicates: has_incident_note, needs_sre_review, passed_unit_tests.\n\nPredicate glossary:\n- has_deployment_approval: has deployment approval\n- has_incident_note: has an incident note\n- may_deploy: may deploy to production\n- needs_sre_review: needs SRE review\n- passed_unit_tests: passed unit tests\n\nCoordinator policy rules:\n- If a service passed unit tests and has deployment approval, then that service may deploy to production.\n\nAgent A report:\nPositive facts:\n- Service C has deployment approval.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Service C passed unit tests.\nNegative facts:\n- none\n\nTarget atom: may_deploy::Service C.\nTarget statement: Service C may deploy to production.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "cloud_deployment_0266_entailed_closed_conclusion__owa__multi_agent", "source_id": "cloud_deployment_0266_entailed_closed_conclusion__owa", "base_id": "cloud_deployment_0266_entailed_closed_conclusion", "split": "train", "domain": "cloud_deployment", "family": "entailed_closed_conclusion", "subset": "control_entailed", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["has_deployment_approval", "has_incident_note", "may_deploy", "needs_sre_review", "passed_unit_tests"], "facts_positive": ["Service C has deployment approval.", "Service C passed unit tests."], "facts_negative": [], "rules_natural": ["If a service passed unit tests and has deployment approval, then that service may deploy to production."], "predicate_glossary": {"passed_unit_tests": "passed unit tests", "has_deployment_approval": "has deployment approval", "has_incident_note": "has an incident note", "may_deploy": "may deploy to production", "needs_sre_review": "needs SRE review"}, "vocabulary_predicates": ["has_deployment_approval", "has_incident_note", "may_deploy", "needs_sre_review", "passed_unit_tests"], "query_atom": "may_deploy::Service C", "target_statement": "Service C may deploy to production.", "gold_reason_type": "entailed_by_fact_or_rule", "symbolic": {"positive_atoms": ["has_deployment_approval::Service C", "passed_unit_tests::Service C"], "negative_atoms": [], "rules": [{"antecedents": ["passed_unit_tests::Service C", "has_deployment_approval::Service C"], "conclusion": "may_deploy::Service C", "text": "If a service passed unit tests and has deployment approval, then that service may deploy to production."}], "query_atom": "may_deploy::Service C", "closure_atoms": ["has_deployment_approval::Service C", "may_deploy::Service C", "passed_unit_tests::Service C"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Service C has deployment approval.", "Service C passed unit tests."], "negative": []}}, "gold_truth_value": "true", "gold_source_used": "agent_b", "gold_closure_handling": "entailed_or_explicit", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: has_deployment_approval, has_incident_note, may_deploy, needs_sre_review, passed_unit_tests.\n\nPredicate glossary:\n- has_deployment_approval: has deployment approval\n- has_incident_note: has an incident note\n- may_deploy: may deploy to production\n- needs_sre_review: needs SRE review\n- passed_unit_tests: passed unit tests\n\nCoordinator policy rules:\n- If a service passed unit tests and has deployment approval, then that service may deploy to production.\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Service C has deployment approval.\n- Service C passed unit tests.\nNegative facts:\n- none\n\nTarget atom: may_deploy::Service C.\nTarget statement: Service C may deploy to production.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "cloud_deployment_0271_entailed_open_conclusion__cwa__multi_agent", "source_id": "cloud_deployment_0271_entailed_open_conclusion__cwa", "base_id": "cloud_deployment_0271_entailed_open_conclusion", "split": "dev", "domain": "cloud_deployment", "family": "entailed_open_conclusion", "subset": "control_entailed", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["has_deployment_approval", "has_incident_note", "may_deploy", "needs_sre_review", "passed_unit_tests"], "agent_a_complete_predicates": ["has_deployment_approval", "has_incident_note", "may_deploy", "needs_sre_review", "passed_unit_tests"], "agent_b_open_predicates": [], "facts_positive": ["Service B has an incident note."], "facts_negative": [], "rules_natural": ["If a service has an incident note, then that service needs SRE review."], "predicate_glossary": {"passed_unit_tests": "passed unit tests", "has_deployment_approval": "has deployment approval", "has_incident_note": "has an incident note", "may_deploy": "may deploy to production", "needs_sre_review": "needs SRE review"}, "vocabulary_predicates": ["has_deployment_approval", "has_incident_note", "may_deploy", "needs_sre_review", "passed_unit_tests"], "query_atom": "needs_sre_review::Service B", "target_statement": "Service B needs SRE review.", "gold_reason_type": "entailed_by_fact_or_rule", "symbolic": {"positive_atoms": ["has_incident_note::Service B"], "negative_atoms": [], "rules": [{"antecedents": ["has_incident_note::Service B"], "conclusion": "needs_sre_review::Service B", "text": "If a service has an incident note, then that service needs SRE review."}], "query_atom": "needs_sre_review::Service B", "closure_atoms": ["has_incident_note::Service B", "needs_sre_review::Service B"]}, "source_reports": {"agent_a": {"positive": ["Service B has an incident note."], "negative": []}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "true", "gold_source_used": "agent_a", "gold_closure_handling": "entailed_or_explicit", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_deployment_approval, has_incident_note, may_deploy, needs_sre_review, passed_unit_tests.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- has_deployment_approval: has deployment approval\n- has_incident_note: has an incident note\n- may_deploy: may deploy to production\n- needs_sre_review: needs SRE review\n- passed_unit_tests: passed unit tests\n\nCoordinator policy rules:\n- If a service has an incident note, then that service needs SRE review.\n\nAgent A report:\nPositive facts:\n- Service B has an incident note.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: needs_sre_review::Service B.\nTarget statement: Service B needs SRE review.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "cloud_deployment_0271_entailed_open_conclusion__lcwa__multi_agent", "source_id": "cloud_deployment_0271_entailed_open_conclusion__lcwa", "base_id": "cloud_deployment_0271_entailed_open_conclusion", "split": "dev", "domain": "cloud_deployment", "family": "entailed_open_conclusion", "subset": "control_entailed", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_deployment_approval", "may_deploy"], "agent_a_complete_predicates": ["has_deployment_approval", "may_deploy"], "agent_b_open_predicates": ["has_incident_note", "needs_sre_review", "passed_unit_tests"], "facts_positive": ["Service B has an incident note."], "facts_negative": [], "rules_natural": ["If a service has an incident note, then that service needs SRE review."], "predicate_glossary": {"passed_unit_tests": "passed unit tests", "has_deployment_approval": "has deployment approval", "has_incident_note": "has an incident note", "may_deploy": "may deploy to production", "needs_sre_review": "needs SRE review"}, "vocabulary_predicates": ["has_deployment_approval", "has_incident_note", "may_deploy", "needs_sre_review", "passed_unit_tests"], "query_atom": "needs_sre_review::Service B", "target_statement": "Service B needs SRE review.", "gold_reason_type": "entailed_by_fact_or_rule", "symbolic": {"positive_atoms": ["has_incident_note::Service B"], "negative_atoms": [], "rules": [{"antecedents": ["has_incident_note::Service B"], "conclusion": "needs_sre_review::Service B", "text": "If a service has an incident note, then that service needs SRE review."}], "query_atom": "needs_sre_review::Service B", "closure_atoms": ["has_incident_note::Service B", "needs_sre_review::Service B"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Service B has an incident note."], "negative": []}}, "gold_truth_value": "true", "gold_source_used": "agent_b", "gold_closure_handling": "entailed_or_explicit", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_deployment_approval, may_deploy.\nAgent B open/incomplete predicates: has_incident_note, needs_sre_review, passed_unit_tests.\n\nPredicate glossary:\n- has_deployment_approval: has deployment approval\n- has_incident_note: has an incident note\n- may_deploy: may deploy to production\n- needs_sre_review: needs SRE review\n- passed_unit_tests: passed unit tests\n\nCoordinator policy rules:\n- If a service has an incident note, then that service needs SRE review.\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Service B has an incident note.\nNegative facts:\n- none\n\nTarget atom: needs_sre_review::Service B.\nTarget statement: Service B needs SRE review.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "cloud_deployment_0271_entailed_open_conclusion__owa__multi_agent", "source_id": "cloud_deployment_0271_entailed_open_conclusion__owa", "base_id": "cloud_deployment_0271_entailed_open_conclusion", "split": "dev", "domain": "cloud_deployment", "family": "entailed_open_conclusion", "subset": "control_entailed", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["has_deployment_approval", "has_incident_note", "may_deploy", "needs_sre_review", "passed_unit_tests"], "facts_positive": ["Service B has an incident note."], "facts_negative": [], "rules_natural": ["If a service has an incident note, then that service needs SRE review."], "predicate_glossary": {"passed_unit_tests": "passed unit tests", "has_deployment_approval": "has deployment approval", "has_incident_note": "has an incident note", "may_deploy": "may deploy to production", "needs_sre_review": "needs SRE review"}, "vocabulary_predicates": ["has_deployment_approval", "has_incident_note", "may_deploy", "needs_sre_review", "passed_unit_tests"], "query_atom": "needs_sre_review::Service B", "target_statement": "Service B needs SRE review.", "gold_reason_type": "entailed_by_fact_or_rule", "symbolic": {"positive_atoms": ["has_incident_note::Service B"], "negative_atoms": [], "rules": [{"antecedents": ["has_incident_note::Service B"], "conclusion": "needs_sre_review::Service B", "text": "If a service has an incident note, then that service needs SRE review."}], "query_atom": "needs_sre_review::Service B", "closure_atoms": ["has_incident_note::Service B", "needs_sre_review::Service B"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Service B has an incident note."], "negative": []}}, "gold_truth_value": "true", "gold_source_used": "agent_b", "gold_closure_handling": "entailed_or_explicit", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: has_deployment_approval, has_incident_note, may_deploy, needs_sre_review, passed_unit_tests.\n\nPredicate glossary:\n- has_deployment_approval: has deployment approval\n- has_incident_note: has an incident note\n- may_deploy: may deploy to production\n- needs_sre_review: needs SRE review\n- passed_unit_tests: passed unit tests\n\nCoordinator policy rules:\n- If a service has an incident note, then that service needs SRE review.\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Service B has an incident note.\nNegative facts:\n- none\n\nTarget atom: needs_sre_review::Service B.\nTarget statement: Service B needs SRE review.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "cloud_deployment_0276_explicit_negative_open__cwa__multi_agent", "source_id": "cloud_deployment_0276_explicit_negative_open__cwa", "base_id": "cloud_deployment_0276_explicit_negative_open", "split": "test", "domain": "cloud_deployment", "family": "explicit_negative_open", "subset": "control_explicit_negative", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["has_deployment_approval", "has_incident_note", "may_deploy", "needs_sre_review", "passed_unit_tests"], "agent_a_complete_predicates": ["has_deployment_approval", "has_incident_note", "may_deploy", "needs_sre_review", "passed_unit_tests"], "agent_b_open_predicates": [], "facts_positive": ["Service C passed unit tests."], "facts_negative": ["Service A does not has an incident note."], "rules_natural": [], "predicate_glossary": {"passed_unit_tests": "passed unit tests", "has_deployment_approval": "has deployment approval", "has_incident_note": "has an incident note", "may_deploy": "may deploy to production", "needs_sre_review": "needs SRE review"}, "vocabulary_predicates": ["has_deployment_approval", "has_incident_note", "may_deploy", "needs_sre_review", "passed_unit_tests"], "query_atom": "has_incident_note::Service A", "target_statement": "Service A has an incident note.", "gold_reason_type": "explicit_negative_fact", "symbolic": {"positive_atoms": ["passed_unit_tests::Service C"], "negative_atoms": ["has_incident_note::Service A"], "rules": [], "query_atom": "has_incident_note::Service A", "closure_atoms": ["passed_unit_tests::Service C"]}, "source_reports": {"agent_a": {"positive": ["Service C passed unit tests."], "negative": ["Service A does not has an incident note."]}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "explicit_negative", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_deployment_approval, has_incident_note, may_deploy, needs_sre_review, passed_unit_tests.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- has_deployment_approval: has deployment approval\n- has_incident_note: has an incident note\n- may_deploy: may deploy to production\n- needs_sre_review: needs SRE review\n- passed_unit_tests: passed unit tests\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- Service C passed unit tests.\nNegative facts:\n- Service A does not has an incident note.\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: has_incident_note::Service A.\nTarget statement: Service A has an incident note.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "cloud_deployment_0276_explicit_negative_open__lcwa__multi_agent", "source_id": "cloud_deployment_0276_explicit_negative_open__lcwa", "base_id": "cloud_deployment_0276_explicit_negative_open", "split": "test", "domain": "cloud_deployment", "family": "explicit_negative_open", "subset": "control_explicit_negative", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_deployment_approval", "may_deploy"], "agent_a_complete_predicates": ["has_deployment_approval", "may_deploy"], "agent_b_open_predicates": ["has_incident_note", "needs_sre_review", "passed_unit_tests"], "facts_positive": ["Service C passed unit tests."], "facts_negative": ["Service A does not has an incident note."], "rules_natural": [], "predicate_glossary": {"passed_unit_tests": "passed unit tests", "has_deployment_approval": "has deployment approval", "has_incident_note": "has an incident note", "may_deploy": "may deploy to production", "needs_sre_review": "needs SRE review"}, "vocabulary_predicates": ["has_deployment_approval", "has_incident_note", "may_deploy", "needs_sre_review", "passed_unit_tests"], "query_atom": "has_incident_note::Service A", "target_statement": "Service A has an incident note.", "gold_reason_type": "explicit_negative_fact", "symbolic": {"positive_atoms": ["passed_unit_tests::Service C"], "negative_atoms": ["has_incident_note::Service A"], "rules": [], "query_atom": "has_incident_note::Service A", "closure_atoms": ["passed_unit_tests::Service C"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Service C passed unit tests."], "negative": ["Service A does not has an incident note."]}}, "gold_truth_value": "false", "gold_source_used": "agent_b", "gold_closure_handling": "explicit_negative", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_deployment_approval, may_deploy.\nAgent B open/incomplete predicates: has_incident_note, needs_sre_review, passed_unit_tests.\n\nPredicate glossary:\n- has_deployment_approval: has deployment approval\n- has_incident_note: has an incident note\n- may_deploy: may deploy to production\n- needs_sre_review: needs SRE review\n- passed_unit_tests: passed unit tests\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Service C passed unit tests.\nNegative facts:\n- Service A does not has an incident note.\n\nTarget atom: has_incident_note::Service A.\nTarget statement: Service A has an incident note.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "cloud_deployment_0276_explicit_negative_open__owa__multi_agent", "source_id": "cloud_deployment_0276_explicit_negative_open__owa", "base_id": "cloud_deployment_0276_explicit_negative_open", "split": "test", "domain": "cloud_deployment", "family": "explicit_negative_open", "subset": "control_explicit_negative", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["has_deployment_approval", "has_incident_note", "may_deploy", "needs_sre_review", "passed_unit_tests"], "facts_positive": ["Service C passed unit tests."], "facts_negative": ["Service A does not has an incident note."], "rules_natural": [], "predicate_glossary": {"passed_unit_tests": "passed unit tests", "has_deployment_approval": "has deployment approval", "has_incident_note": "has an incident note", "may_deploy": "may deploy to production", "needs_sre_review": "needs SRE review"}, "vocabulary_predicates": ["has_deployment_approval", "has_incident_note", "may_deploy", "needs_sre_review", "passed_unit_tests"], "query_atom": "has_incident_note::Service A", "target_statement": "Service A has an incident note.", "gold_reason_type": "explicit_negative_fact", "symbolic": {"positive_atoms": ["passed_unit_tests::Service C"], "negative_atoms": ["has_incident_note::Service A"], "rules": [], "query_atom": "has_incident_note::Service A", "closure_atoms": ["passed_unit_tests::Service C"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Service C passed unit tests."], "negative": ["Service A does not has an incident note."]}}, "gold_truth_value": "false", "gold_source_used": "agent_b", "gold_closure_handling": "explicit_negative", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: has_deployment_approval, has_incident_note, may_deploy, needs_sre_review, passed_unit_tests.\n\nPredicate glossary:\n- has_deployment_approval: has deployment approval\n- has_incident_note: has an incident note\n- may_deploy: may deploy to production\n- needs_sre_review: needs SRE review\n- passed_unit_tests: passed unit tests\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Service C passed unit tests.\nNegative facts:\n- Service A does not has an incident note.\n\nTarget atom: has_incident_note::Service A.\nTarget statement: Service A has an incident note.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "finance_controls_0120_closed_missing_direct__cwa__multi_agent", "source_id": "finance_controls_0120_closed_missing_direct__cwa", "base_id": "finance_controls_0120_closed_missing_direct", "split": "train", "domain": "finance_controls", "family": "closed_missing_direct", "subset": "core_contrastive", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["has_compliance_clearance", "has_manual_exception", "may_receive_wire", "needs_compliance_review", "passed_kyc"], "agent_a_complete_predicates": ["has_compliance_clearance", "has_manual_exception", "may_receive_wire", "needs_compliance_review", "passed_kyc"], "agent_b_open_predicates": [], "facts_positive": ["Account 12 passed KYC checks."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_kyc": "passed KYC checks", "has_compliance_clearance": "has compliance clearance", "has_manual_exception": "has a manual exception", "may_receive_wire": "may receive a wire transfer", "needs_compliance_review": "needs compliance review"}, "vocabulary_predicates": ["has_compliance_clearance", "has_manual_exception", "may_receive_wire", "needs_compliance_review", "passed_kyc"], "query_atom": "has_compliance_clearance::Account 12", "target_statement": "Account 12 has compliance clearance.", "gold_reason_type": "global_closed_world_underivable", "symbolic": {"positive_atoms": ["passed_kyc::Account 12"], "negative_atoms": [], "rules": [], "query_atom": "has_compliance_clearance::Account 12", "closure_atoms": ["passed_kyc::Account 12"]}, "source_reports": {"agent_a": {"positive": ["Account 12 passed KYC checks."], "negative": []}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "closed_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_compliance_clearance, has_manual_exception, may_receive_wire, needs_compliance_review, passed_kyc.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- has_compliance_clearance: has compliance clearance\n- has_manual_exception: has a manual exception\n- may_receive_wire: may receive a wire transfer\n- needs_compliance_review: needs compliance review\n- passed_kyc: passed KYC checks\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- Account 12 passed KYC checks.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: has_compliance_clearance::Account 12.\nTarget statement: Account 12 has compliance clearance.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "finance_controls_0120_closed_missing_direct__lcwa__multi_agent", "source_id": "finance_controls_0120_closed_missing_direct__lcwa", "base_id": "finance_controls_0120_closed_missing_direct", "split": "train", "domain": "finance_controls", "family": "closed_missing_direct", "subset": "core_contrastive", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_compliance_clearance"], "agent_a_complete_predicates": ["has_compliance_clearance"], "agent_b_open_predicates": ["has_manual_exception", "may_receive_wire", "needs_compliance_review", "passed_kyc"], "facts_positive": ["Account 12 passed KYC checks."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_kyc": "passed KYC checks", "has_compliance_clearance": "has compliance clearance", "has_manual_exception": "has a manual exception", "may_receive_wire": "may receive a wire transfer", "needs_compliance_review": "needs compliance review"}, "vocabulary_predicates": ["has_compliance_clearance", "has_manual_exception", "may_receive_wire", "needs_compliance_review", "passed_kyc"], "query_atom": "has_compliance_clearance::Account 12", "target_statement": "Account 12 has compliance clearance.", "gold_reason_type": "local_closed_world_underivable", "symbolic": {"positive_atoms": ["passed_kyc::Account 12"], "negative_atoms": [], "rules": [], "query_atom": "has_compliance_clearance::Account 12", "closure_atoms": ["passed_kyc::Account 12"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Account 12 passed KYC checks."], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "closed_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_compliance_clearance.\nAgent B open/incomplete predicates: has_manual_exception, may_receive_wire, needs_compliance_review, passed_kyc.\n\nPredicate glossary:\n- has_compliance_clearance: has compliance clearance\n- has_manual_exception: has a manual exception\n- may_receive_wire: may receive a wire transfer\n- needs_compliance_review: needs compliance review\n- passed_kyc: passed KYC checks\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Account 12 passed KYC checks.\nNegative facts:\n- none\n\nTarget atom: has_compliance_clearance::Account 12.\nTarget statement: Account 12 has compliance clearance.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "finance_controls_0120_closed_missing_direct__owa__multi_agent", "source_id": "finance_controls_0120_closed_missing_direct__owa", "base_id": "finance_controls_0120_closed_missing_direct", "split": "train", "domain": "finance_controls", "family": "closed_missing_direct", "subset": "core_contrastive", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["has_compliance_clearance", "has_manual_exception", "may_receive_wire", "needs_compliance_review", "passed_kyc"], "facts_positive": ["Account 12 passed KYC checks."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_kyc": "passed KYC checks", "has_compliance_clearance": "has compliance clearance", "has_manual_exception": "has a manual exception", "may_receive_wire": "may receive a wire transfer", "needs_compliance_review": "needs compliance review"}, "vocabulary_predicates": ["has_compliance_clearance", "has_manual_exception", "may_receive_wire", "needs_compliance_review", "passed_kyc"], "query_atom": "has_compliance_clearance::Account 12", "target_statement": "Account 12 has compliance clearance.", "gold_reason_type": "open_world_underivable", "symbolic": {"positive_atoms": ["passed_kyc::Account 12"], "negative_atoms": [], "rules": [], "query_atom": "has_compliance_clearance::Account 12", "closure_atoms": ["passed_kyc::Account 12"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Account 12 passed KYC checks."], "negative": []}}, "gold_truth_value": "unknown", "gold_source_used": "agent_b", "gold_closure_handling": "open_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: has_compliance_clearance, has_manual_exception, may_receive_wire, needs_compliance_review, passed_kyc.\n\nPredicate glossary:\n- has_compliance_clearance: has compliance clearance\n- has_manual_exception: has a manual exception\n- may_receive_wire: may receive a wire transfer\n- needs_compliance_review: needs compliance review\n- passed_kyc: passed KYC checks\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Account 12 passed KYC checks.\nNegative facts:\n- none\n\nTarget atom: has_compliance_clearance::Account 12.\nTarget statement: Account 12 has compliance clearance.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "finance_controls_0121_closed_missing_direct__cwa__multi_agent", "source_id": "finance_controls_0121_closed_missing_direct__cwa", "base_id": "finance_controls_0121_closed_missing_direct", "split": "train", "domain": "finance_controls", "family": "closed_missing_direct", "subset": "core_contrastive", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["has_compliance_clearance", "has_manual_exception", "may_receive_wire", "needs_compliance_review", "passed_kyc"], "agent_a_complete_predicates": ["has_compliance_clearance", "has_manual_exception", "may_receive_wire", "needs_compliance_review", "passed_kyc"], "agent_b_open_predicates": [], "facts_positive": ["Account 27 passed KYC checks."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_kyc": "passed KYC checks", "has_compliance_clearance": "has compliance clearance", "has_manual_exception": "has a manual exception", "may_receive_wire": "may receive a wire transfer", "needs_compliance_review": "needs compliance review"}, "vocabulary_predicates": ["has_compliance_clearance", "has_manual_exception", "may_receive_wire", "needs_compliance_review", "passed_kyc"], "query_atom": "has_compliance_clearance::Account 27", "target_statement": "Account 27 has compliance clearance.", "gold_reason_type": "global_closed_world_underivable", "symbolic": {"positive_atoms": ["passed_kyc::Account 27"], "negative_atoms": [], "rules": [], "query_atom": "has_compliance_clearance::Account 27", "closure_atoms": ["passed_kyc::Account 27"]}, "source_reports": {"agent_a": {"positive": ["Account 27 passed KYC checks."], "negative": []}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "closed_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_compliance_clearance, has_manual_exception, may_receive_wire, needs_compliance_review, passed_kyc.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- has_compliance_clearance: has compliance clearance\n- has_manual_exception: has a manual exception\n- may_receive_wire: may receive a wire transfer\n- needs_compliance_review: needs compliance review\n- passed_kyc: passed KYC checks\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- Account 27 passed KYC checks.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: has_compliance_clearance::Account 27.\nTarget statement: Account 27 has compliance clearance.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "finance_controls_0121_closed_missing_direct__lcwa__multi_agent", "source_id": "finance_controls_0121_closed_missing_direct__lcwa", "base_id": "finance_controls_0121_closed_missing_direct", "split": "train", "domain": "finance_controls", "family": "closed_missing_direct", "subset": "core_contrastive", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_compliance_clearance"], "agent_a_complete_predicates": ["has_compliance_clearance"], "agent_b_open_predicates": ["has_manual_exception", "may_receive_wire", "needs_compliance_review", "passed_kyc"], "facts_positive": ["Account 27 passed KYC checks."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_kyc": "passed KYC checks", "has_compliance_clearance": "has compliance clearance", "has_manual_exception": "has a manual exception", "may_receive_wire": "may receive a wire transfer", "needs_compliance_review": "needs compliance review"}, "vocabulary_predicates": ["has_compliance_clearance", "has_manual_exception", "may_receive_wire", "needs_compliance_review", "passed_kyc"], "query_atom": "has_compliance_clearance::Account 27", "target_statement": "Account 27 has compliance clearance.", "gold_reason_type": "local_closed_world_underivable", "symbolic": {"positive_atoms": ["passed_kyc::Account 27"], "negative_atoms": [], "rules": [], "query_atom": "has_compliance_clearance::Account 27", "closure_atoms": ["passed_kyc::Account 27"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Account 27 passed KYC checks."], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "closed_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_compliance_clearance.\nAgent B open/incomplete predicates: has_manual_exception, may_receive_wire, needs_compliance_review, passed_kyc.\n\nPredicate glossary:\n- has_compliance_clearance: has compliance clearance\n- has_manual_exception: has a manual exception\n- may_receive_wire: may receive a wire transfer\n- needs_compliance_review: needs compliance review\n- passed_kyc: passed KYC checks\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Account 27 passed KYC checks.\nNegative facts:\n- none\n\nTarget atom: has_compliance_clearance::Account 27.\nTarget statement: Account 27 has compliance clearance.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "finance_controls_0121_closed_missing_direct__owa__multi_agent", "source_id": "finance_controls_0121_closed_missing_direct__owa", "base_id": "finance_controls_0121_closed_missing_direct", "split": "train", "domain": "finance_controls", "family": "closed_missing_direct", "subset": "core_contrastive", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["has_compliance_clearance", "has_manual_exception", "may_receive_wire", "needs_compliance_review", "passed_kyc"], "facts_positive": ["Account 27 passed KYC checks."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_kyc": "passed KYC checks", "has_compliance_clearance": "has compliance clearance", "has_manual_exception": "has a manual exception", "may_receive_wire": "may receive a wire transfer", "needs_compliance_review": "needs compliance review"}, "vocabulary_predicates": ["has_compliance_clearance", "has_manual_exception", "may_receive_wire", "needs_compliance_review", "passed_kyc"], "query_atom": "has_compliance_clearance::Account 27", "target_statement": "Account 27 has compliance clearance.", "gold_reason_type": "open_world_underivable", "symbolic": {"positive_atoms": ["passed_kyc::Account 27"], "negative_atoms": [], "rules": [], "query_atom": "has_compliance_clearance::Account 27", "closure_atoms": ["passed_kyc::Account 27"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Account 27 passed KYC checks."], "negative": []}}, "gold_truth_value": "unknown", "gold_source_used": "agent_b", "gold_closure_handling": "open_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: has_compliance_clearance, has_manual_exception, may_receive_wire, needs_compliance_review, passed_kyc.\n\nPredicate glossary:\n- has_compliance_clearance: has compliance clearance\n- has_manual_exception: has a manual exception\n- may_receive_wire: may receive a wire transfer\n- needs_compliance_review: needs compliance review\n- passed_kyc: passed KYC checks\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Account 27 passed KYC checks.\nNegative facts:\n- none\n\nTarget atom: has_compliance_clearance::Account 27.\nTarget statement: Account 27 has compliance clearance.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "finance_controls_0122_closed_missing_direct__cwa__multi_agent", "source_id": "finance_controls_0122_closed_missing_direct__cwa", "base_id": "finance_controls_0122_closed_missing_direct", "split": "train", "domain": "finance_controls", "family": "closed_missing_direct", "subset": "core_contrastive", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["has_compliance_clearance", "has_manual_exception", "may_receive_wire", "needs_compliance_review", "passed_kyc"], "agent_a_complete_predicates": ["has_compliance_clearance", "has_manual_exception", "may_receive_wire", "needs_compliance_review", "passed_kyc"], "agent_b_open_predicates": [], "facts_positive": ["Account 35 passed KYC checks."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_kyc": "passed KYC checks", "has_compliance_clearance": "has compliance clearance", "has_manual_exception": "has a manual exception", "may_receive_wire": "may receive a wire transfer", "needs_compliance_review": "needs compliance review"}, "vocabulary_predicates": ["has_compliance_clearance", "has_manual_exception", "may_receive_wire", "needs_compliance_review", "passed_kyc"], "query_atom": "has_compliance_clearance::Account 35", "target_statement": "Account 35 has compliance clearance.", "gold_reason_type": "global_closed_world_underivable", "symbolic": {"positive_atoms": ["passed_kyc::Account 35"], "negative_atoms": [], "rules": [], "query_atom": "has_compliance_clearance::Account 35", "closure_atoms": ["passed_kyc::Account 35"]}, "source_reports": {"agent_a": {"positive": ["Account 35 passed KYC checks."], "negative": []}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "closed_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_compliance_clearance, has_manual_exception, may_receive_wire, needs_compliance_review, passed_kyc.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- has_compliance_clearance: has compliance clearance\n- has_manual_exception: has a manual exception\n- may_receive_wire: may receive a wire transfer\n- needs_compliance_review: needs compliance review\n- passed_kyc: passed KYC checks\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- Account 35 passed KYC checks.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: has_compliance_clearance::Account 35.\nTarget statement: Account 35 has compliance clearance.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "finance_controls_0122_closed_missing_direct__lcwa__multi_agent", "source_id": "finance_controls_0122_closed_missing_direct__lcwa", "base_id": "finance_controls_0122_closed_missing_direct", "split": "train", "domain": "finance_controls", "family": "closed_missing_direct", "subset": "core_contrastive", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_compliance_clearance"], "agent_a_complete_predicates": ["has_compliance_clearance"], "agent_b_open_predicates": ["has_manual_exception", "may_receive_wire", "needs_compliance_review", "passed_kyc"], "facts_positive": ["Account 35 passed KYC checks."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_kyc": "passed KYC checks", "has_compliance_clearance": "has compliance clearance", "has_manual_exception": "has a manual exception", "may_receive_wire": "may receive a wire transfer", "needs_compliance_review": "needs compliance review"}, "vocabulary_predicates": ["has_compliance_clearance", "has_manual_exception", "may_receive_wire", "needs_compliance_review", "passed_kyc"], "query_atom": "has_compliance_clearance::Account 35", "target_statement": "Account 35 has compliance clearance.", "gold_reason_type": "local_closed_world_underivable", "symbolic": {"positive_atoms": ["passed_kyc::Account 35"], "negative_atoms": [], "rules": [], "query_atom": "has_compliance_clearance::Account 35", "closure_atoms": ["passed_kyc::Account 35"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Account 35 passed KYC checks."], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "closed_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_compliance_clearance.\nAgent B open/incomplete predicates: has_manual_exception, may_receive_wire, needs_compliance_review, passed_kyc.\n\nPredicate glossary:\n- has_compliance_clearance: has compliance clearance\n- has_manual_exception: has a manual exception\n- may_receive_wire: may receive a wire transfer\n- needs_compliance_review: needs compliance review\n- passed_kyc: passed KYC checks\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Account 35 passed KYC checks.\nNegative facts:\n- none\n\nTarget atom: has_compliance_clearance::Account 35.\nTarget statement: Account 35 has compliance clearance.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "finance_controls_0122_closed_missing_direct__owa__multi_agent", "source_id": "finance_controls_0122_closed_missing_direct__owa", "base_id": "finance_controls_0122_closed_missing_direct", "split": "train", "domain": "finance_controls", "family": "closed_missing_direct", "subset": "core_contrastive", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["has_compliance_clearance", "has_manual_exception", "may_receive_wire", "needs_compliance_review", "passed_kyc"], "facts_positive": ["Account 35 passed KYC checks."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_kyc": "passed KYC checks", "has_compliance_clearance": "has compliance clearance", "has_manual_exception": "has a manual exception", "may_receive_wire": "may receive a wire transfer", "needs_compliance_review": "needs compliance review"}, "vocabulary_predicates": ["has_compliance_clearance", "has_manual_exception", "may_receive_wire", "needs_compliance_review", "passed_kyc"], "query_atom": "has_compliance_clearance::Account 35", "target_statement": "Account 35 has compliance clearance.", "gold_reason_type": "open_world_underivable", "symbolic": {"positive_atoms": ["passed_kyc::Account 35"], "negative_atoms": [], "rules": [], "query_atom": "has_compliance_clearance::Account 35", "closure_atoms": ["passed_kyc::Account 35"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Account 35 passed KYC checks."], "negative": []}}, "gold_truth_value": "unknown", "gold_source_used": "agent_b", "gold_closure_handling": "open_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: has_compliance_clearance, has_manual_exception, may_receive_wire, needs_compliance_review, passed_kyc.\n\nPredicate glossary:\n- has_compliance_clearance: has compliance clearance\n- has_manual_exception: has a manual exception\n- may_receive_wire: may receive a wire transfer\n- needs_compliance_review: needs compliance review\n- passed_kyc: passed KYC checks\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Account 35 passed KYC checks.\nNegative facts:\n- none\n\nTarget atom: has_compliance_clearance::Account 35.\nTarget statement: Account 35 has compliance clearance.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "finance_controls_0123_closed_missing_direct__cwa__multi_agent", "source_id": "finance_controls_0123_closed_missing_direct__cwa", "base_id": "finance_controls_0123_closed_missing_direct", "split": "test", "domain": "finance_controls", "family": "closed_missing_direct", "subset": "core_contrastive", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["has_compliance_clearance", "has_manual_exception", "may_receive_wire", "needs_compliance_review", "passed_kyc"], "agent_a_complete_predicates": ["has_compliance_clearance", "has_manual_exception", "may_receive_wire", "needs_compliance_review", "passed_kyc"], "agent_b_open_predicates": [], "facts_positive": ["Account 48 passed KYC checks."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_kyc": "passed KYC checks", "has_compliance_clearance": "has compliance clearance", "has_manual_exception": "has a manual exception", "may_receive_wire": "may receive a wire transfer", "needs_compliance_review": "needs compliance review"}, "vocabulary_predicates": ["has_compliance_clearance", "has_manual_exception", "may_receive_wire", "needs_compliance_review", "passed_kyc"], "query_atom": "has_compliance_clearance::Account 48", "target_statement": "Account 48 has compliance clearance.", "gold_reason_type": "global_closed_world_underivable", "symbolic": {"positive_atoms": ["passed_kyc::Account 48"], "negative_atoms": [], "rules": [], "query_atom": "has_compliance_clearance::Account 48", "closure_atoms": ["passed_kyc::Account 48"]}, "source_reports": {"agent_a": {"positive": ["Account 48 passed KYC checks."], "negative": []}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "closed_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_compliance_clearance, has_manual_exception, may_receive_wire, needs_compliance_review, passed_kyc.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- has_compliance_clearance: has compliance clearance\n- has_manual_exception: has a manual exception\n- may_receive_wire: may receive a wire transfer\n- needs_compliance_review: needs compliance review\n- passed_kyc: passed KYC checks\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- Account 48 passed KYC checks.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: has_compliance_clearance::Account 48.\nTarget statement: Account 48 has compliance clearance.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "finance_controls_0123_closed_missing_direct__lcwa__multi_agent", "source_id": "finance_controls_0123_closed_missing_direct__lcwa", "base_id": "finance_controls_0123_closed_missing_direct", "split": "test", "domain": "finance_controls", "family": "closed_missing_direct", "subset": "core_contrastive", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_compliance_clearance"], "agent_a_complete_predicates": ["has_compliance_clearance"], "agent_b_open_predicates": ["has_manual_exception", "may_receive_wire", "needs_compliance_review", "passed_kyc"], "facts_positive": ["Account 48 passed KYC checks."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_kyc": "passed KYC checks", "has_compliance_clearance": "has compliance clearance", "has_manual_exception": "has a manual exception", "may_receive_wire": "may receive a wire transfer", "needs_compliance_review": "needs compliance review"}, "vocabulary_predicates": ["has_compliance_clearance", "has_manual_exception", "may_receive_wire", "needs_compliance_review", "passed_kyc"], "query_atom": "has_compliance_clearance::Account 48", "target_statement": "Account 48 has compliance clearance.", "gold_reason_type": "local_closed_world_underivable", "symbolic": {"positive_atoms": ["passed_kyc::Account 48"], "negative_atoms": [], "rules": [], "query_atom": "has_compliance_clearance::Account 48", "closure_atoms": ["passed_kyc::Account 48"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Account 48 passed KYC checks."], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "closed_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_compliance_clearance.\nAgent B open/incomplete predicates: has_manual_exception, may_receive_wire, needs_compliance_review, passed_kyc.\n\nPredicate glossary:\n- has_compliance_clearance: has compliance clearance\n- has_manual_exception: has a manual exception\n- may_receive_wire: may receive a wire transfer\n- needs_compliance_review: needs compliance review\n- passed_kyc: passed KYC checks\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Account 48 passed KYC checks.\nNegative facts:\n- none\n\nTarget atom: has_compliance_clearance::Account 48.\nTarget statement: Account 48 has compliance clearance.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "finance_controls_0123_closed_missing_direct__owa__multi_agent", "source_id": "finance_controls_0123_closed_missing_direct__owa", "base_id": "finance_controls_0123_closed_missing_direct", "split": "test", "domain": "finance_controls", "family": "closed_missing_direct", "subset": "core_contrastive", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["has_compliance_clearance", "has_manual_exception", "may_receive_wire", "needs_compliance_review", "passed_kyc"], "facts_positive": ["Account 48 passed KYC checks."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_kyc": "passed KYC checks", "has_compliance_clearance": "has compliance clearance", "has_manual_exception": "has a manual exception", "may_receive_wire": "may receive a wire transfer", "needs_compliance_review": "needs compliance review"}, "vocabulary_predicates": ["has_compliance_clearance", "has_manual_exception", "may_receive_wire", "needs_compliance_review", "passed_kyc"], "query_atom": "has_compliance_clearance::Account 48", "target_statement": "Account 48 has compliance clearance.", "gold_reason_type": "open_world_underivable", "symbolic": {"positive_atoms": ["passed_kyc::Account 48"], "negative_atoms": [], "rules": [], "query_atom": "has_compliance_clearance::Account 48", "closure_atoms": ["passed_kyc::Account 48"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Account 48 passed KYC checks."], "negative": []}}, "gold_truth_value": "unknown", "gold_source_used": "agent_b", "gold_closure_handling": "open_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: has_compliance_clearance, has_manual_exception, may_receive_wire, needs_compliance_review, passed_kyc.\n\nPredicate glossary:\n- has_compliance_clearance: has compliance clearance\n- has_manual_exception: has a manual exception\n- may_receive_wire: may receive a wire transfer\n- needs_compliance_review: needs compliance review\n- passed_kyc: passed KYC checks\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Account 48 passed KYC checks.\nNegative facts:\n- none\n\nTarget atom: has_compliance_clearance::Account 48.\nTarget statement: Account 48 has compliance clearance.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "finance_controls_0124_open_missing_direct__cwa__multi_agent", "source_id": "finance_controls_0124_open_missing_direct__cwa", "base_id": "finance_controls_0124_open_missing_direct", "split": "test", "domain": "finance_controls", "family": "open_missing_direct", "subset": "core_contrastive", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["has_compliance_clearance", "has_manual_exception", "may_receive_wire", "needs_compliance_review", "passed_kyc"], "agent_a_complete_predicates": ["has_compliance_clearance", "has_manual_exception", "may_receive_wire", "needs_compliance_review", "passed_kyc"], "agent_b_open_predicates": [], "facts_positive": ["Account 59 passed KYC checks."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_kyc": "passed KYC checks", "has_compliance_clearance": "has compliance clearance", "has_manual_exception": "has a manual exception", "may_receive_wire": "may receive a wire transfer", "needs_compliance_review": "needs compliance review"}, "vocabulary_predicates": ["has_compliance_clearance", "has_manual_exception", "may_receive_wire", "needs_compliance_review", "passed_kyc"], "query_atom": "has_manual_exception::Account 59", "target_statement": "Account 59 has a manual exception.", "gold_reason_type": "global_closed_world_underivable", "symbolic": {"positive_atoms": ["passed_kyc::Account 59"], "negative_atoms": [], "rules": [], "query_atom": "has_manual_exception::Account 59", "closure_atoms": ["passed_kyc::Account 59"]}, "source_reports": {"agent_a": {"positive": ["Account 59 passed KYC checks."], "negative": []}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "closed_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_compliance_clearance, has_manual_exception, may_receive_wire, needs_compliance_review, passed_kyc.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- has_compliance_clearance: has compliance clearance\n- has_manual_exception: has a manual exception\n- may_receive_wire: may receive a wire transfer\n- needs_compliance_review: needs compliance review\n- passed_kyc: passed KYC checks\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- Account 59 passed KYC checks.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: has_manual_exception::Account 59.\nTarget statement: Account 59 has a manual exception.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "finance_controls_0124_open_missing_direct__lcwa__multi_agent", "source_id": "finance_controls_0124_open_missing_direct__lcwa", "base_id": "finance_controls_0124_open_missing_direct", "split": "test", "domain": "finance_controls", "family": "open_missing_direct", "subset": "core_contrastive", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_compliance_clearance", "may_receive_wire"], "agent_a_complete_predicates": ["has_compliance_clearance", "may_receive_wire"], "agent_b_open_predicates": ["has_manual_exception", "needs_compliance_review", "passed_kyc"], "facts_positive": ["Account 59 passed KYC checks."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_kyc": "passed KYC checks", "has_compliance_clearance": "has compliance clearance", "has_manual_exception": "has a manual exception", "may_receive_wire": "may receive a wire transfer", "needs_compliance_review": "needs compliance review"}, "vocabulary_predicates": ["has_compliance_clearance", "has_manual_exception", "may_receive_wire", "needs_compliance_review", "passed_kyc"], "query_atom": "has_manual_exception::Account 59", "target_statement": "Account 59 has a manual exception.", "gold_reason_type": "local_open_world_underivable", "symbolic": {"positive_atoms": ["passed_kyc::Account 59"], "negative_atoms": [], "rules": [], "query_atom": "has_manual_exception::Account 59", "closure_atoms": ["passed_kyc::Account 59"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Account 59 passed KYC checks."], "negative": []}}, "gold_truth_value": "unknown", "gold_source_used": "agent_b", "gold_closure_handling": "open_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_compliance_clearance, may_receive_wire.\nAgent B open/incomplete predicates: has_manual_exception, needs_compliance_review, passed_kyc.\n\nPredicate glossary:\n- has_compliance_clearance: has compliance clearance\n- has_manual_exception: has a manual exception\n- may_receive_wire: may receive a wire transfer\n- needs_compliance_review: needs compliance review\n- passed_kyc: passed KYC checks\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Account 59 passed KYC checks.\nNegative facts:\n- none\n\nTarget atom: has_manual_exception::Account 59.\nTarget statement: Account 59 has a manual exception.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "finance_controls_0124_open_missing_direct__owa__multi_agent", "source_id": "finance_controls_0124_open_missing_direct__owa", "base_id": "finance_controls_0124_open_missing_direct", "split": "test", "domain": "finance_controls", "family": "open_missing_direct", "subset": "core_contrastive", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["has_compliance_clearance", "has_manual_exception", "may_receive_wire", "needs_compliance_review", "passed_kyc"], "facts_positive": ["Account 59 passed KYC checks."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_kyc": "passed KYC checks", "has_compliance_clearance": "has compliance clearance", "has_manual_exception": "has a manual exception", "may_receive_wire": "may receive a wire transfer", "needs_compliance_review": "needs compliance review"}, "vocabulary_predicates": ["has_compliance_clearance", "has_manual_exception", "may_receive_wire", "needs_compliance_review", "passed_kyc"], "query_atom": "has_manual_exception::Account 59", "target_statement": "Account 59 has a manual exception.", "gold_reason_type": "open_world_underivable", "symbolic": {"positive_atoms": ["passed_kyc::Account 59"], "negative_atoms": [], "rules": [], "query_atom": "has_manual_exception::Account 59", "closure_atoms": ["passed_kyc::Account 59"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Account 59 passed KYC checks."], "negative": []}}, "gold_truth_value": "unknown", "gold_source_used": "agent_b", "gold_closure_handling": "open_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: has_compliance_clearance, has_manual_exception, may_receive_wire, needs_compliance_review, passed_kyc.\n\nPredicate glossary:\n- has_compliance_clearance: has compliance clearance\n- has_manual_exception: has a manual exception\n- may_receive_wire: may receive a wire transfer\n- needs_compliance_review: needs compliance review\n- passed_kyc: passed KYC checks\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Account 59 passed KYC checks.\nNegative facts:\n- none\n\nTarget atom: has_manual_exception::Account 59.\nTarget statement: Account 59 has a manual exception.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "finance_controls_0125_open_missing_direct__cwa__multi_agent", "source_id": "finance_controls_0125_open_missing_direct__cwa", "base_id": "finance_controls_0125_open_missing_direct", "split": "train", "domain": "finance_controls", "family": "open_missing_direct", "subset": "core_contrastive", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["has_compliance_clearance", "has_manual_exception", "may_receive_wire", "needs_compliance_review", "passed_kyc"], "agent_a_complete_predicates": ["has_compliance_clearance", "has_manual_exception", "may_receive_wire", "needs_compliance_review", "passed_kyc"], "agent_b_open_predicates": [], "facts_positive": ["Account 61 passed KYC checks."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_kyc": "passed KYC checks", "has_compliance_clearance": "has compliance clearance", "has_manual_exception": "has a manual exception", "may_receive_wire": "may receive a wire transfer", "needs_compliance_review": "needs compliance review"}, "vocabulary_predicates": ["has_compliance_clearance", "has_manual_exception", "may_receive_wire", "needs_compliance_review", "passed_kyc"], "query_atom": "has_manual_exception::Account 61", "target_statement": "Account 61 has a manual exception.", "gold_reason_type": "global_closed_world_underivable", "symbolic": {"positive_atoms": ["passed_kyc::Account 61"], "negative_atoms": [], "rules": [], "query_atom": "has_manual_exception::Account 61", "closure_atoms": ["passed_kyc::Account 61"]}, "source_reports": {"agent_a": {"positive": ["Account 61 passed KYC checks."], "negative": []}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "closed_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_compliance_clearance, has_manual_exception, may_receive_wire, needs_compliance_review, passed_kyc.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- has_compliance_clearance: has compliance clearance\n- has_manual_exception: has a manual exception\n- may_receive_wire: may receive a wire transfer\n- needs_compliance_review: needs compliance review\n- passed_kyc: passed KYC checks\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- Account 61 passed KYC checks.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: has_manual_exception::Account 61.\nTarget statement: Account 61 has a manual exception.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "finance_controls_0125_open_missing_direct__lcwa__multi_agent", "source_id": "finance_controls_0125_open_missing_direct__lcwa", "base_id": "finance_controls_0125_open_missing_direct", "split": "train", "domain": "finance_controls", "family": "open_missing_direct", "subset": "core_contrastive", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_compliance_clearance", "may_receive_wire"], "agent_a_complete_predicates": ["has_compliance_clearance", "may_receive_wire"], "agent_b_open_predicates": ["has_manual_exception", "needs_compliance_review", "passed_kyc"], "facts_positive": ["Account 61 passed KYC checks."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_kyc": "passed KYC checks", "has_compliance_clearance": "has compliance clearance", "has_manual_exception": "has a manual exception", "may_receive_wire": "may receive a wire transfer", "needs_compliance_review": "needs compliance review"}, "vocabulary_predicates": ["has_compliance_clearance", "has_manual_exception", "may_receive_wire", "needs_compliance_review", "passed_kyc"], "query_atom": "has_manual_exception::Account 61", "target_statement": "Account 61 has a manual exception.", "gold_reason_type": "local_open_world_underivable", "symbolic": {"positive_atoms": ["passed_kyc::Account 61"], "negative_atoms": [], "rules": [], "query_atom": "has_manual_exception::Account 61", "closure_atoms": ["passed_kyc::Account 61"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Account 61 passed KYC checks."], "negative": []}}, "gold_truth_value": "unknown", "gold_source_used": "agent_b", "gold_closure_handling": "open_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_compliance_clearance, may_receive_wire.\nAgent B open/incomplete predicates: has_manual_exception, needs_compliance_review, passed_kyc.\n\nPredicate glossary:\n- has_compliance_clearance: has compliance clearance\n- has_manual_exception: has a manual exception\n- may_receive_wire: may receive a wire transfer\n- needs_compliance_review: needs compliance review\n- passed_kyc: passed KYC checks\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Account 61 passed KYC checks.\nNegative facts:\n- none\n\nTarget atom: has_manual_exception::Account 61.\nTarget statement: Account 61 has a manual exception.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "finance_controls_0125_open_missing_direct__owa__multi_agent", "source_id": "finance_controls_0125_open_missing_direct__owa", "base_id": "finance_controls_0125_open_missing_direct", "split": "train", "domain": "finance_controls", "family": "open_missing_direct", "subset": "core_contrastive", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["has_compliance_clearance", "has_manual_exception", "may_receive_wire", "needs_compliance_review", "passed_kyc"], "facts_positive": ["Account 61 passed KYC checks."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_kyc": "passed KYC checks", "has_compliance_clearance": "has compliance clearance", "has_manual_exception": "has a manual exception", "may_receive_wire": "may receive a wire transfer", "needs_compliance_review": "needs compliance review"}, "vocabulary_predicates": ["has_compliance_clearance", "has_manual_exception", "may_receive_wire", "needs_compliance_review", "passed_kyc"], "query_atom": "has_manual_exception::Account 61", "target_statement": "Account 61 has a manual exception.", "gold_reason_type": "open_world_underivable", "symbolic": {"positive_atoms": ["passed_kyc::Account 61"], "negative_atoms": [], "rules": [], "query_atom": "has_manual_exception::Account 61", "closure_atoms": ["passed_kyc::Account 61"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Account 61 passed KYC checks."], "negative": []}}, "gold_truth_value": "unknown", "gold_source_used": "agent_b", "gold_closure_handling": "open_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: has_compliance_clearance, has_manual_exception, may_receive_wire, needs_compliance_review, passed_kyc.\n\nPredicate glossary:\n- has_compliance_clearance: has compliance clearance\n- has_manual_exception: has a manual exception\n- may_receive_wire: may receive a wire transfer\n- needs_compliance_review: needs compliance review\n- passed_kyc: passed KYC checks\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Account 61 passed KYC checks.\nNegative facts:\n- none\n\nTarget atom: has_manual_exception::Account 61.\nTarget statement: Account 61 has a manual exception.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "finance_controls_0129_closed_missing_with_open_distractor__cwa__multi_agent", "source_id": "finance_controls_0129_closed_missing_with_open_distractor__cwa", "base_id": "finance_controls_0129_closed_missing_with_open_distractor", "split": "train", "domain": "finance_controls", "family": "closed_missing_with_open_distractor", "subset": "core_contrastive", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["has_compliance_clearance", "has_manual_exception", "may_receive_wire", "needs_compliance_review", "passed_kyc"], "agent_a_complete_predicates": ["has_compliance_clearance", "has_manual_exception", "may_receive_wire", "needs_compliance_review", "passed_kyc"], "agent_b_open_predicates": [], "facts_positive": ["Account 59 has a manual exception.", "Account 48 passed KYC checks."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_kyc": "passed KYC checks", "has_compliance_clearance": "has compliance clearance", "has_manual_exception": "has a manual exception", "may_receive_wire": "may receive a wire transfer", "needs_compliance_review": "needs compliance review"}, "vocabulary_predicates": ["has_compliance_clearance", "has_manual_exception", "may_receive_wire", "needs_compliance_review", "passed_kyc"], "query_atom": "has_compliance_clearance::Account 48", "target_statement": "Account 48 has compliance clearance.", "gold_reason_type": "global_closed_world_underivable", "symbolic": {"positive_atoms": ["has_manual_exception::Account 59", "passed_kyc::Account 48"], "negative_atoms": [], "rules": [], "query_atom": "has_compliance_clearance::Account 48", "closure_atoms": ["has_manual_exception::Account 59", "passed_kyc::Account 48"]}, "source_reports": {"agent_a": {"positive": ["Account 59 has a manual exception.", "Account 48 passed KYC checks."], "negative": []}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "closed_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_compliance_clearance, has_manual_exception, may_receive_wire, needs_compliance_review, passed_kyc.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- has_compliance_clearance: has compliance clearance\n- has_manual_exception: has a manual exception\n- may_receive_wire: may receive a wire transfer\n- needs_compliance_review: needs compliance review\n- passed_kyc: passed KYC checks\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- Account 59 has a manual exception.\n- Account 48 passed KYC checks.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: has_compliance_clearance::Account 48.\nTarget statement: Account 48 has compliance clearance.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "finance_controls_0129_closed_missing_with_open_distractor__lcwa__multi_agent", "source_id": "finance_controls_0129_closed_missing_with_open_distractor__lcwa", "base_id": "finance_controls_0129_closed_missing_with_open_distractor", "split": "train", "domain": "finance_controls", "family": "closed_missing_with_open_distractor", "subset": "core_contrastive", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_compliance_clearance"], "agent_a_complete_predicates": ["has_compliance_clearance"], "agent_b_open_predicates": ["has_manual_exception", "may_receive_wire", "needs_compliance_review", "passed_kyc"], "facts_positive": ["Account 59 has a manual exception.", "Account 48 passed KYC checks."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_kyc": "passed KYC checks", "has_compliance_clearance": "has compliance clearance", "has_manual_exception": "has a manual exception", "may_receive_wire": "may receive a wire transfer", "needs_compliance_review": "needs compliance review"}, "vocabulary_predicates": ["has_compliance_clearance", "has_manual_exception", "may_receive_wire", "needs_compliance_review", "passed_kyc"], "query_atom": "has_compliance_clearance::Account 48", "target_statement": "Account 48 has compliance clearance.", "gold_reason_type": "local_closed_world_underivable", "symbolic": {"positive_atoms": ["has_manual_exception::Account 59", "passed_kyc::Account 48"], "negative_atoms": [], "rules": [], "query_atom": "has_compliance_clearance::Account 48", "closure_atoms": ["has_manual_exception::Account 59", "passed_kyc::Account 48"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Account 59 has a manual exception.", "Account 48 passed KYC checks."], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "closed_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_compliance_clearance.\nAgent B open/incomplete predicates: has_manual_exception, may_receive_wire, needs_compliance_review, passed_kyc.\n\nPredicate glossary:\n- has_compliance_clearance: has compliance clearance\n- has_manual_exception: has a manual exception\n- may_receive_wire: may receive a wire transfer\n- needs_compliance_review: needs compliance review\n- passed_kyc: passed KYC checks\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Account 59 has a manual exception.\n- Account 48 passed KYC checks.\nNegative facts:\n- none\n\nTarget atom: has_compliance_clearance::Account 48.\nTarget statement: Account 48 has compliance clearance.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "finance_controls_0129_closed_missing_with_open_distractor__owa__multi_agent", "source_id": "finance_controls_0129_closed_missing_with_open_distractor__owa", "base_id": "finance_controls_0129_closed_missing_with_open_distractor", "split": "train", "domain": "finance_controls", "family": "closed_missing_with_open_distractor", "subset": "core_contrastive", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["has_compliance_clearance", "has_manual_exception", "may_receive_wire", "needs_compliance_review", "passed_kyc"], "facts_positive": ["Account 59 has a manual exception.", "Account 48 passed KYC checks."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_kyc": "passed KYC checks", "has_compliance_clearance": "has compliance clearance", "has_manual_exception": "has a manual exception", "may_receive_wire": "may receive a wire transfer", "needs_compliance_review": "needs compliance review"}, "vocabulary_predicates": ["has_compliance_clearance", "has_manual_exception", "may_receive_wire", "needs_compliance_review", "passed_kyc"], "query_atom": "has_compliance_clearance::Account 48", "target_statement": "Account 48 has compliance clearance.", "gold_reason_type": "open_world_underivable", "symbolic": {"positive_atoms": ["has_manual_exception::Account 59", "passed_kyc::Account 48"], "negative_atoms": [], "rules": [], "query_atom": "has_compliance_clearance::Account 48", "closure_atoms": ["has_manual_exception::Account 59", "passed_kyc::Account 48"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Account 59 has a manual exception.", "Account 48 passed KYC checks."], "negative": []}}, "gold_truth_value": "unknown", "gold_source_used": "agent_b", "gold_closure_handling": "open_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: has_compliance_clearance, has_manual_exception, may_receive_wire, needs_compliance_review, passed_kyc.\n\nPredicate glossary:\n- has_compliance_clearance: has compliance clearance\n- has_manual_exception: has a manual exception\n- may_receive_wire: may receive a wire transfer\n- needs_compliance_review: needs compliance review\n- passed_kyc: passed KYC checks\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Account 59 has a manual exception.\n- Account 48 passed KYC checks.\nNegative facts:\n- none\n\nTarget atom: has_compliance_clearance::Account 48.\nTarget statement: Account 48 has compliance clearance.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "finance_controls_0132_open_missing_with_closed_distractor__cwa__multi_agent", "source_id": "finance_controls_0132_open_missing_with_closed_distractor__cwa", "base_id": "finance_controls_0132_open_missing_with_closed_distractor", "split": "train", "domain": "finance_controls", "family": "open_missing_with_closed_distractor", "subset": "core_contrastive", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["has_compliance_clearance", "has_manual_exception", "may_receive_wire", "needs_compliance_review", "passed_kyc"], "agent_a_complete_predicates": ["has_compliance_clearance", "has_manual_exception", "may_receive_wire", "needs_compliance_review", "passed_kyc"], "agent_b_open_predicates": [], "facts_positive": ["Account 27 has compliance clearance.", "Account 27 passed KYC checks."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_kyc": "passed KYC checks", "has_compliance_clearance": "has compliance clearance", "has_manual_exception": "has a manual exception", "may_receive_wire": "may receive a wire transfer", "needs_compliance_review": "needs compliance review"}, "vocabulary_predicates": ["has_compliance_clearance", "has_manual_exception", "may_receive_wire", "needs_compliance_review", "passed_kyc"], "query_atom": "has_manual_exception::Account 12", "target_statement": "Account 12 has a manual exception.", "gold_reason_type": "global_closed_world_underivable", "symbolic": {"positive_atoms": ["has_compliance_clearance::Account 27", "passed_kyc::Account 27"], "negative_atoms": [], "rules": [], "query_atom": "has_manual_exception::Account 12", "closure_atoms": ["has_compliance_clearance::Account 27", "passed_kyc::Account 27"]}, "source_reports": {"agent_a": {"positive": ["Account 27 has compliance clearance.", "Account 27 passed KYC checks."], "negative": []}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "closed_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_compliance_clearance, has_manual_exception, may_receive_wire, needs_compliance_review, passed_kyc.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- has_compliance_clearance: has compliance clearance\n- has_manual_exception: has a manual exception\n- may_receive_wire: may receive a wire transfer\n- needs_compliance_review: needs compliance review\n- passed_kyc: passed KYC checks\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- Account 27 has compliance clearance.\n- Account 27 passed KYC checks.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: has_manual_exception::Account 12.\nTarget statement: Account 12 has a manual exception.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "finance_controls_0132_open_missing_with_closed_distractor__lcwa__multi_agent", "source_id": "finance_controls_0132_open_missing_with_closed_distractor__lcwa", "base_id": "finance_controls_0132_open_missing_with_closed_distractor", "split": "train", "domain": "finance_controls", "family": "open_missing_with_closed_distractor", "subset": "core_contrastive", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_compliance_clearance", "may_receive_wire"], "agent_a_complete_predicates": ["has_compliance_clearance", "may_receive_wire"], "agent_b_open_predicates": ["has_manual_exception", "needs_compliance_review", "passed_kyc"], "facts_positive": ["Account 27 has compliance clearance.", "Account 27 passed KYC checks."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_kyc": "passed KYC checks", "has_compliance_clearance": "has compliance clearance", "has_manual_exception": "has a manual exception", "may_receive_wire": "may receive a wire transfer", "needs_compliance_review": "needs compliance review"}, "vocabulary_predicates": ["has_compliance_clearance", "has_manual_exception", "may_receive_wire", "needs_compliance_review", "passed_kyc"], "query_atom": "has_manual_exception::Account 12", "target_statement": "Account 12 has a manual exception.", "gold_reason_type": "local_open_world_underivable", "symbolic": {"positive_atoms": ["has_compliance_clearance::Account 27", "passed_kyc::Account 27"], "negative_atoms": [], "rules": [], "query_atom": "has_manual_exception::Account 12", "closure_atoms": ["has_compliance_clearance::Account 27", "passed_kyc::Account 27"]}, "source_reports": {"agent_a": {"positive": ["Account 27 has compliance clearance."], "negative": []}, "agent_b": {"positive": ["Account 27 passed KYC checks."], "negative": []}}, "gold_truth_value": "unknown", "gold_source_used": "agent_b", "gold_closure_handling": "open_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_compliance_clearance, may_receive_wire.\nAgent B open/incomplete predicates: has_manual_exception, needs_compliance_review, passed_kyc.\n\nPredicate glossary:\n- has_compliance_clearance: has compliance clearance\n- has_manual_exception: has a manual exception\n- may_receive_wire: may receive a wire transfer\n- needs_compliance_review: needs compliance review\n- passed_kyc: passed KYC checks\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- Account 27 has compliance clearance.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Account 27 passed KYC checks.\nNegative facts:\n- none\n\nTarget atom: has_manual_exception::Account 12.\nTarget statement: Account 12 has a manual exception.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "finance_controls_0132_open_missing_with_closed_distractor__owa__multi_agent", "source_id": "finance_controls_0132_open_missing_with_closed_distractor__owa", "base_id": "finance_controls_0132_open_missing_with_closed_distractor", "split": "train", "domain": "finance_controls", "family": "open_missing_with_closed_distractor", "subset": "core_contrastive", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["has_compliance_clearance", "has_manual_exception", "may_receive_wire", "needs_compliance_review", "passed_kyc"], "facts_positive": ["Account 27 has compliance clearance.", "Account 27 passed KYC checks."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_kyc": "passed KYC checks", "has_compliance_clearance": "has compliance clearance", "has_manual_exception": "has a manual exception", "may_receive_wire": "may receive a wire transfer", "needs_compliance_review": "needs compliance review"}, "vocabulary_predicates": ["has_compliance_clearance", "has_manual_exception", "may_receive_wire", "needs_compliance_review", "passed_kyc"], "query_atom": "has_manual_exception::Account 12", "target_statement": "Account 12 has a manual exception.", "gold_reason_type": "open_world_underivable", "symbolic": {"positive_atoms": ["has_compliance_clearance::Account 27", "passed_kyc::Account 27"], "negative_atoms": [], "rules": [], "query_atom": "has_manual_exception::Account 12", "closure_atoms": ["has_compliance_clearance::Account 27", "passed_kyc::Account 27"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Account 27 has compliance clearance.", "Account 27 passed KYC checks."], "negative": []}}, "gold_truth_value": "unknown", "gold_source_used": "agent_b", "gold_closure_handling": "open_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: has_compliance_clearance, has_manual_exception, may_receive_wire, needs_compliance_review, passed_kyc.\n\nPredicate glossary:\n- has_compliance_clearance: has compliance clearance\n- has_manual_exception: has a manual exception\n- may_receive_wire: may receive a wire transfer\n- needs_compliance_review: needs compliance review\n- passed_kyc: passed KYC checks\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Account 27 has compliance clearance.\n- Account 27 passed KYC checks.\nNegative facts:\n- none\n\nTarget atom: has_manual_exception::Account 12.\nTarget statement: Account 12 has a manual exception.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "finance_controls_0133_open_missing_with_closed_distractor__cwa__multi_agent", "source_id": "finance_controls_0133_open_missing_with_closed_distractor__cwa", "base_id": "finance_controls_0133_open_missing_with_closed_distractor", "split": "train", "domain": "finance_controls", "family": "open_missing_with_closed_distractor", "subset": "core_contrastive", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["has_compliance_clearance", "has_manual_exception", "may_receive_wire", "needs_compliance_review", "passed_kyc"], "agent_a_complete_predicates": ["has_compliance_clearance", "has_manual_exception", "may_receive_wire", "needs_compliance_review", "passed_kyc"], "agent_b_open_predicates": [], "facts_positive": ["Account 35 has compliance clearance.", "Account 35 passed KYC checks."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_kyc": "passed KYC checks", "has_compliance_clearance": "has compliance clearance", "has_manual_exception": "has a manual exception", "may_receive_wire": "may receive a wire transfer", "needs_compliance_review": "needs compliance review"}, "vocabulary_predicates": ["has_compliance_clearance", "has_manual_exception", "may_receive_wire", "needs_compliance_review", "passed_kyc"], "query_atom": "has_manual_exception::Account 27", "target_statement": "Account 27 has a manual exception.", "gold_reason_type": "global_closed_world_underivable", "symbolic": {"positive_atoms": ["has_compliance_clearance::Account 35", "passed_kyc::Account 35"], "negative_atoms": [], "rules": [], "query_atom": "has_manual_exception::Account 27", "closure_atoms": ["has_compliance_clearance::Account 35", "passed_kyc::Account 35"]}, "source_reports": {"agent_a": {"positive": ["Account 35 has compliance clearance.", "Account 35 passed KYC checks."], "negative": []}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "closed_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_compliance_clearance, has_manual_exception, may_receive_wire, needs_compliance_review, passed_kyc.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- has_compliance_clearance: has compliance clearance\n- has_manual_exception: has a manual exception\n- may_receive_wire: may receive a wire transfer\n- needs_compliance_review: needs compliance review\n- passed_kyc: passed KYC checks\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- Account 35 has compliance clearance.\n- Account 35 passed KYC checks.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: has_manual_exception::Account 27.\nTarget statement: Account 27 has a manual exception.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "finance_controls_0133_open_missing_with_closed_distractor__lcwa__multi_agent", "source_id": "finance_controls_0133_open_missing_with_closed_distractor__lcwa", "base_id": "finance_controls_0133_open_missing_with_closed_distractor", "split": "train", "domain": "finance_controls", "family": "open_missing_with_closed_distractor", "subset": "core_contrastive", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_compliance_clearance", "may_receive_wire"], "agent_a_complete_predicates": ["has_compliance_clearance", "may_receive_wire"], "agent_b_open_predicates": ["has_manual_exception", "needs_compliance_review", "passed_kyc"], "facts_positive": ["Account 35 has compliance clearance.", "Account 35 passed KYC checks."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_kyc": "passed KYC checks", "has_compliance_clearance": "has compliance clearance", "has_manual_exception": "has a manual exception", "may_receive_wire": "may receive a wire transfer", "needs_compliance_review": "needs compliance review"}, "vocabulary_predicates": ["has_compliance_clearance", "has_manual_exception", "may_receive_wire", "needs_compliance_review", "passed_kyc"], "query_atom": "has_manual_exception::Account 27", "target_statement": "Account 27 has a manual exception.", "gold_reason_type": "local_open_world_underivable", "symbolic": {"positive_atoms": ["has_compliance_clearance::Account 35", "passed_kyc::Account 35"], "negative_atoms": [], "rules": [], "query_atom": "has_manual_exception::Account 27", "closure_atoms": ["has_compliance_clearance::Account 35", "passed_kyc::Account 35"]}, "source_reports": {"agent_a": {"positive": ["Account 35 has compliance clearance."], "negative": []}, "agent_b": {"positive": ["Account 35 passed KYC checks."], "negative": []}}, "gold_truth_value": "unknown", "gold_source_used": "agent_b", "gold_closure_handling": "open_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_compliance_clearance, may_receive_wire.\nAgent B open/incomplete predicates: has_manual_exception, needs_compliance_review, passed_kyc.\n\nPredicate glossary:\n- has_compliance_clearance: has compliance clearance\n- has_manual_exception: has a manual exception\n- may_receive_wire: may receive a wire transfer\n- needs_compliance_review: needs compliance review\n- passed_kyc: passed KYC checks\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- Account 35 has compliance clearance.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Account 35 passed KYC checks.\nNegative facts:\n- none\n\nTarget atom: has_manual_exception::Account 27.\nTarget statement: Account 27 has a manual exception.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "finance_controls_0133_open_missing_with_closed_distractor__owa__multi_agent", "source_id": "finance_controls_0133_open_missing_with_closed_distractor__owa", "base_id": "finance_controls_0133_open_missing_with_closed_distractor", "split": "train", "domain": "finance_controls", "family": "open_missing_with_closed_distractor", "subset": "core_contrastive", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["has_compliance_clearance", "has_manual_exception", "may_receive_wire", "needs_compliance_review", "passed_kyc"], "facts_positive": ["Account 35 has compliance clearance.", "Account 35 passed KYC checks."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_kyc": "passed KYC checks", "has_compliance_clearance": "has compliance clearance", "has_manual_exception": "has a manual exception", "may_receive_wire": "may receive a wire transfer", "needs_compliance_review": "needs compliance review"}, "vocabulary_predicates": ["has_compliance_clearance", "has_manual_exception", "may_receive_wire", "needs_compliance_review", "passed_kyc"], "query_atom": "has_manual_exception::Account 27", "target_statement": "Account 27 has a manual exception.", "gold_reason_type": "open_world_underivable", "symbolic": {"positive_atoms": ["has_compliance_clearance::Account 35", "passed_kyc::Account 35"], "negative_atoms": [], "rules": [], "query_atom": "has_manual_exception::Account 27", "closure_atoms": ["has_compliance_clearance::Account 35", "passed_kyc::Account 35"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Account 35 has compliance clearance.", "Account 35 passed KYC checks."], "negative": []}}, "gold_truth_value": "unknown", "gold_source_used": "agent_b", "gold_closure_handling": "open_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: has_compliance_clearance, has_manual_exception, may_receive_wire, needs_compliance_review, passed_kyc.\n\nPredicate glossary:\n- has_compliance_clearance: has compliance clearance\n- has_manual_exception: has a manual exception\n- may_receive_wire: may receive a wire transfer\n- needs_compliance_review: needs compliance review\n- passed_kyc: passed KYC checks\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Account 35 has compliance clearance.\n- Account 35 passed KYC checks.\nNegative facts:\n- none\n\nTarget atom: has_manual_exception::Account 27.\nTarget statement: Account 27 has a manual exception.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "finance_controls_0139_closed_derived_missing_antecedent__cwa__multi_agent", "source_id": "finance_controls_0139_closed_derived_missing_antecedent__cwa", "base_id": "finance_controls_0139_closed_derived_missing_antecedent", "split": "train", "domain": "finance_controls", "family": "closed_derived_missing_antecedent", "subset": "core_contrastive", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["has_compliance_clearance", "has_manual_exception", "may_receive_wire", "needs_compliance_review", "passed_kyc"], "agent_a_complete_predicates": ["has_compliance_clearance", "has_manual_exception", "may_receive_wire", "needs_compliance_review", "passed_kyc"], "agent_b_open_predicates": [], "facts_positive": ["Account 27 passed KYC checks."], "facts_negative": [], "rules_natural": ["If an account passed KYC checks and has compliance clearance, then that account may receive a wire transfer."], "predicate_glossary": {"passed_kyc": "passed KYC checks", "has_compliance_clearance": "has compliance clearance", "has_manual_exception": "has a manual exception", "may_receive_wire": "may receive a wire transfer", "needs_compliance_review": "needs compliance review"}, "vocabulary_predicates": ["has_compliance_clearance", "has_manual_exception", "may_receive_wire", "needs_compliance_review", "passed_kyc"], "query_atom": "may_receive_wire::Account 27", "target_statement": "Account 27 may receive a wire transfer.", "gold_reason_type": "global_closed_world_underivable", "symbolic": {"positive_atoms": ["passed_kyc::Account 27"], "negative_atoms": [], "rules": [{"antecedents": ["passed_kyc::Account 27", "has_compliance_clearance::Account 27"], "conclusion": "may_receive_wire::Account 27", "text": "If an account passed KYC checks and has compliance clearance, then that account may receive a wire transfer."}], "query_atom": "may_receive_wire::Account 27", "closure_atoms": ["passed_kyc::Account 27"]}, "source_reports": {"agent_a": {"positive": ["Account 27 passed KYC checks."], "negative": []}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "closed_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_compliance_clearance, has_manual_exception, may_receive_wire, needs_compliance_review, passed_kyc.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- has_compliance_clearance: has compliance clearance\n- has_manual_exception: has a manual exception\n- may_receive_wire: may receive a wire transfer\n- needs_compliance_review: needs compliance review\n- passed_kyc: passed KYC checks\n\nCoordinator policy rules:\n- If an account passed KYC checks and has compliance clearance, then that account may receive a wire transfer.\n\nAgent A report:\nPositive facts:\n- Account 27 passed KYC checks.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: may_receive_wire::Account 27.\nTarget statement: Account 27 may receive a wire transfer.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "finance_controls_0139_closed_derived_missing_antecedent__lcwa__multi_agent", "source_id": "finance_controls_0139_closed_derived_missing_antecedent__lcwa", "base_id": "finance_controls_0139_closed_derived_missing_antecedent", "split": "train", "domain": "finance_controls", "family": "closed_derived_missing_antecedent", "subset": "core_contrastive", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_compliance_clearance", "may_receive_wire"], "agent_a_complete_predicates": ["has_compliance_clearance", "may_receive_wire"], "agent_b_open_predicates": ["has_manual_exception", "needs_compliance_review", "passed_kyc"], "facts_positive": ["Account 27 passed KYC checks."], "facts_negative": [], "rules_natural": ["If an account passed KYC checks and has compliance clearance, then that account may receive a wire transfer."], "predicate_glossary": {"passed_kyc": "passed KYC checks", "has_compliance_clearance": "has compliance clearance", "has_manual_exception": "has a manual exception", "may_receive_wire": "may receive a wire transfer", "needs_compliance_review": "needs compliance review"}, "vocabulary_predicates": ["has_compliance_clearance", "has_manual_exception", "may_receive_wire", "needs_compliance_review", "passed_kyc"], "query_atom": "may_receive_wire::Account 27", "target_statement": "Account 27 may receive a wire transfer.", "gold_reason_type": "local_closed_world_underivable", "symbolic": {"positive_atoms": ["passed_kyc::Account 27"], "negative_atoms": [], "rules": [{"antecedents": ["passed_kyc::Account 27", "has_compliance_clearance::Account 27"], "conclusion": "may_receive_wire::Account 27", "text": "If an account passed KYC checks and has compliance clearance, then that account may receive a wire transfer."}], "query_atom": "may_receive_wire::Account 27", "closure_atoms": ["passed_kyc::Account 27"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Account 27 passed KYC checks."], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "closed_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_compliance_clearance, may_receive_wire.\nAgent B open/incomplete predicates: has_manual_exception, needs_compliance_review, passed_kyc.\n\nPredicate glossary:\n- has_compliance_clearance: has compliance clearance\n- has_manual_exception: has a manual exception\n- may_receive_wire: may receive a wire transfer\n- needs_compliance_review: needs compliance review\n- passed_kyc: passed KYC checks\n\nCoordinator policy rules:\n- If an account passed KYC checks and has compliance clearance, then that account may receive a wire transfer.\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Account 27 passed KYC checks.\nNegative facts:\n- none\n\nTarget atom: may_receive_wire::Account 27.\nTarget statement: Account 27 may receive a wire transfer.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "finance_controls_0139_closed_derived_missing_antecedent__owa__multi_agent", "source_id": "finance_controls_0139_closed_derived_missing_antecedent__owa", "base_id": "finance_controls_0139_closed_derived_missing_antecedent", "split": "train", "domain": "finance_controls", "family": "closed_derived_missing_antecedent", "subset": "core_contrastive", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["has_compliance_clearance", "has_manual_exception", "may_receive_wire", "needs_compliance_review", "passed_kyc"], "facts_positive": ["Account 27 passed KYC checks."], "facts_negative": [], "rules_natural": ["If an account passed KYC checks and has compliance clearance, then that account may receive a wire transfer."], "predicate_glossary": {"passed_kyc": "passed KYC checks", "has_compliance_clearance": "has compliance clearance", "has_manual_exception": "has a manual exception", "may_receive_wire": "may receive a wire transfer", "needs_compliance_review": "needs compliance review"}, "vocabulary_predicates": ["has_compliance_clearance", "has_manual_exception", "may_receive_wire", "needs_compliance_review", "passed_kyc"], "query_atom": "may_receive_wire::Account 27", "target_statement": "Account 27 may receive a wire transfer.", "gold_reason_type": "open_world_underivable", "symbolic": {"positive_atoms": ["passed_kyc::Account 27"], "negative_atoms": [], "rules": [{"antecedents": ["passed_kyc::Account 27", "has_compliance_clearance::Account 27"], "conclusion": "may_receive_wire::Account 27", "text": "If an account passed KYC checks and has compliance clearance, then that account may receive a wire transfer."}], "query_atom": "may_receive_wire::Account 27", "closure_atoms": ["passed_kyc::Account 27"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Account 27 passed KYC checks."], "negative": []}}, "gold_truth_value": "unknown", "gold_source_used": "agent_b", "gold_closure_handling": "open_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: has_compliance_clearance, has_manual_exception, may_receive_wire, needs_compliance_review, passed_kyc.\n\nPredicate glossary:\n- has_compliance_clearance: has compliance clearance\n- has_manual_exception: has a manual exception\n- may_receive_wire: may receive a wire transfer\n- needs_compliance_review: needs compliance review\n- passed_kyc: passed KYC checks\n\nCoordinator policy rules:\n- If an account passed KYC checks and has compliance clearance, then that account may receive a wire transfer.\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Account 27 passed KYC checks.\nNegative facts:\n- none\n\nTarget atom: may_receive_wire::Account 27.\nTarget statement: Account 27 may receive a wire transfer.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "finance_controls_0140_open_derived_missing_antecedent__cwa__multi_agent", "source_id": "finance_controls_0140_open_derived_missing_antecedent__cwa", "base_id": "finance_controls_0140_open_derived_missing_antecedent", "split": "train", "domain": "finance_controls", "family": "open_derived_missing_antecedent", "subset": "core_contrastive", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["has_compliance_clearance", "has_manual_exception", "may_receive_wire", "needs_compliance_review", "passed_kyc"], "agent_a_complete_predicates": ["has_compliance_clearance", "has_manual_exception", "may_receive_wire", "needs_compliance_review", "passed_kyc"], "agent_b_open_predicates": [], "facts_positive": ["Account 35 passed KYC checks."], "facts_negative": [], "rules_natural": ["If an account has a manual exception, then that account needs compliance review."], "predicate_glossary": {"passed_kyc": "passed KYC checks", "has_compliance_clearance": "has compliance clearance", "has_manual_exception": "has a manual exception", "may_receive_wire": "may receive a wire transfer", "needs_compliance_review": "needs compliance review"}, "vocabulary_predicates": ["has_compliance_clearance", "has_manual_exception", "may_receive_wire", "needs_compliance_review", "passed_kyc"], "query_atom": "needs_compliance_review::Account 35", "target_statement": "Account 35 needs compliance review.", "gold_reason_type": "global_closed_world_underivable", "symbolic": {"positive_atoms": ["passed_kyc::Account 35"], "negative_atoms": [], "rules": [{"antecedents": ["has_manual_exception::Account 35"], "conclusion": "needs_compliance_review::Account 35", "text": "If an account has a manual exception, then that account needs compliance review."}], "query_atom": "needs_compliance_review::Account 35", "closure_atoms": ["passed_kyc::Account 35"]}, "source_reports": {"agent_a": {"positive": ["Account 35 passed KYC checks."], "negative": []}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "closed_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_compliance_clearance, has_manual_exception, may_receive_wire, needs_compliance_review, passed_kyc.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- has_compliance_clearance: has compliance clearance\n- has_manual_exception: has a manual exception\n- may_receive_wire: may receive a wire transfer\n- needs_compliance_review: needs compliance review\n- passed_kyc: passed KYC checks\n\nCoordinator policy rules:\n- If an account has a manual exception, then that account needs compliance review.\n\nAgent A report:\nPositive facts:\n- Account 35 passed KYC checks.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: needs_compliance_review::Account 35.\nTarget statement: Account 35 needs compliance review.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "finance_controls_0140_open_derived_missing_antecedent__lcwa__multi_agent", "source_id": "finance_controls_0140_open_derived_missing_antecedent__lcwa", "base_id": "finance_controls_0140_open_derived_missing_antecedent", "split": "train", "domain": "finance_controls", "family": "open_derived_missing_antecedent", "subset": "core_contrastive", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_compliance_clearance", "may_receive_wire"], "agent_a_complete_predicates": ["has_compliance_clearance", "may_receive_wire"], "agent_b_open_predicates": ["has_manual_exception", "needs_compliance_review", "passed_kyc"], "facts_positive": ["Account 35 passed KYC checks."], "facts_negative": [], "rules_natural": ["If an account has a manual exception, then that account needs compliance review."], "predicate_glossary": {"passed_kyc": "passed KYC checks", "has_compliance_clearance": "has compliance clearance", "has_manual_exception": "has a manual exception", "may_receive_wire": "may receive a wire transfer", "needs_compliance_review": "needs compliance review"}, "vocabulary_predicates": ["has_compliance_clearance", "has_manual_exception", "may_receive_wire", "needs_compliance_review", "passed_kyc"], "query_atom": "needs_compliance_review::Account 35", "target_statement": "Account 35 needs compliance review.", "gold_reason_type": "local_open_world_underivable", "symbolic": {"positive_atoms": ["passed_kyc::Account 35"], "negative_atoms": [], "rules": [{"antecedents": ["has_manual_exception::Account 35"], "conclusion": "needs_compliance_review::Account 35", "text": "If an account has a manual exception, then that account needs compliance review."}], "query_atom": "needs_compliance_review::Account 35", "closure_atoms": ["passed_kyc::Account 35"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Account 35 passed KYC checks."], "negative": []}}, "gold_truth_value": "unknown", "gold_source_used": "agent_b", "gold_closure_handling": "open_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_compliance_clearance, may_receive_wire.\nAgent B open/incomplete predicates: has_manual_exception, needs_compliance_review, passed_kyc.\n\nPredicate glossary:\n- has_compliance_clearance: has compliance clearance\n- has_manual_exception: has a manual exception\n- may_receive_wire: may receive a wire transfer\n- needs_compliance_review: needs compliance review\n- passed_kyc: passed KYC checks\n\nCoordinator policy rules:\n- If an account has a manual exception, then that account needs compliance review.\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Account 35 passed KYC checks.\nNegative facts:\n- none\n\nTarget atom: needs_compliance_review::Account 35.\nTarget statement: Account 35 needs compliance review.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "finance_controls_0140_open_derived_missing_antecedent__owa__multi_agent", "source_id": "finance_controls_0140_open_derived_missing_antecedent__owa", "base_id": "finance_controls_0140_open_derived_missing_antecedent", "split": "train", "domain": "finance_controls", "family": "open_derived_missing_antecedent", "subset": "core_contrastive", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["has_compliance_clearance", "has_manual_exception", "may_receive_wire", "needs_compliance_review", "passed_kyc"], "facts_positive": ["Account 35 passed KYC checks."], "facts_negative": [], "rules_natural": ["If an account has a manual exception, then that account needs compliance review."], "predicate_glossary": {"passed_kyc": "passed KYC checks", "has_compliance_clearance": "has compliance clearance", "has_manual_exception": "has a manual exception", "may_receive_wire": "may receive a wire transfer", "needs_compliance_review": "needs compliance review"}, "vocabulary_predicates": ["has_compliance_clearance", "has_manual_exception", "may_receive_wire", "needs_compliance_review", "passed_kyc"], "query_atom": "needs_compliance_review::Account 35", "target_statement": "Account 35 needs compliance review.", "gold_reason_type": "open_world_underivable", "symbolic": {"positive_atoms": ["passed_kyc::Account 35"], "negative_atoms": [], "rules": [{"antecedents": ["has_manual_exception::Account 35"], "conclusion": "needs_compliance_review::Account 35", "text": "If an account has a manual exception, then that account needs compliance review."}], "query_atom": "needs_compliance_review::Account 35", "closure_atoms": ["passed_kyc::Account 35"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Account 35 passed KYC checks."], "negative": []}}, "gold_truth_value": "unknown", "gold_source_used": "agent_b", "gold_closure_handling": "open_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: has_compliance_clearance, has_manual_exception, may_receive_wire, needs_compliance_review, passed_kyc.\n\nPredicate glossary:\n- has_compliance_clearance: has compliance clearance\n- has_manual_exception: has a manual exception\n- may_receive_wire: may receive a wire transfer\n- needs_compliance_review: needs compliance review\n- passed_kyc: passed KYC checks\n\nCoordinator policy rules:\n- If an account has a manual exception, then that account needs compliance review.\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Account 35 passed KYC checks.\nNegative facts:\n- none\n\nTarget atom: needs_compliance_review::Account 35.\nTarget statement: Account 35 needs compliance review.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "finance_controls_0144_entailed_closed_conclusion__cwa__multi_agent", "source_id": "finance_controls_0144_entailed_closed_conclusion__cwa", "base_id": "finance_controls_0144_entailed_closed_conclusion", "split": "train", "domain": "finance_controls", "family": "entailed_closed_conclusion", "subset": "control_entailed", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["has_compliance_clearance", "has_manual_exception", "may_receive_wire", "needs_compliance_review", "passed_kyc"], "agent_a_complete_predicates": ["has_compliance_clearance", "has_manual_exception", "may_receive_wire", "needs_compliance_review", "passed_kyc"], "agent_b_open_predicates": [], "facts_positive": ["Account 12 has compliance clearance.", "Account 12 passed KYC checks."], "facts_negative": [], "rules_natural": ["If an account passed KYC checks and has compliance clearance, then that account may receive a wire transfer."], "predicate_glossary": {"passed_kyc": "passed KYC checks", "has_compliance_clearance": "has compliance clearance", "has_manual_exception": "has a manual exception", "may_receive_wire": "may receive a wire transfer", "needs_compliance_review": "needs compliance review"}, "vocabulary_predicates": ["has_compliance_clearance", "has_manual_exception", "may_receive_wire", "needs_compliance_review", "passed_kyc"], "query_atom": "may_receive_wire::Account 12", "target_statement": "Account 12 may receive a wire transfer.", "gold_reason_type": "entailed_by_fact_or_rule", "symbolic": {"positive_atoms": ["has_compliance_clearance::Account 12", "passed_kyc::Account 12"], "negative_atoms": [], "rules": [{"antecedents": ["passed_kyc::Account 12", "has_compliance_clearance::Account 12"], "conclusion": "may_receive_wire::Account 12", "text": "If an account passed KYC checks and has compliance clearance, then that account may receive a wire transfer."}], "query_atom": "may_receive_wire::Account 12", "closure_atoms": ["has_compliance_clearance::Account 12", "may_receive_wire::Account 12", "passed_kyc::Account 12"]}, "source_reports": {"agent_a": {"positive": ["Account 12 has compliance clearance.", "Account 12 passed KYC checks."], "negative": []}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "true", "gold_source_used": "agent_a", "gold_closure_handling": "entailed_or_explicit", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_compliance_clearance, has_manual_exception, may_receive_wire, needs_compliance_review, passed_kyc.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- has_compliance_clearance: has compliance clearance\n- has_manual_exception: has a manual exception\n- may_receive_wire: may receive a wire transfer\n- needs_compliance_review: needs compliance review\n- passed_kyc: passed KYC checks\n\nCoordinator policy rules:\n- If an account passed KYC checks and has compliance clearance, then that account may receive a wire transfer.\n\nAgent A report:\nPositive facts:\n- Account 12 has compliance clearance.\n- Account 12 passed KYC checks.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: may_receive_wire::Account 12.\nTarget statement: Account 12 may receive a wire transfer.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "finance_controls_0144_entailed_closed_conclusion__lcwa__multi_agent", "source_id": "finance_controls_0144_entailed_closed_conclusion__lcwa", "base_id": "finance_controls_0144_entailed_closed_conclusion", "split": "train", "domain": "finance_controls", "family": "entailed_closed_conclusion", "subset": "control_entailed", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_compliance_clearance", "may_receive_wire"], "agent_a_complete_predicates": ["has_compliance_clearance", "may_receive_wire"], "agent_b_open_predicates": ["has_manual_exception", "needs_compliance_review", "passed_kyc"], "facts_positive": ["Account 12 has compliance clearance.", "Account 12 passed KYC checks."], "facts_negative": [], "rules_natural": ["If an account passed KYC checks and has compliance clearance, then that account may receive a wire transfer."], "predicate_glossary": {"passed_kyc": "passed KYC checks", "has_compliance_clearance": "has compliance clearance", "has_manual_exception": "has a manual exception", "may_receive_wire": "may receive a wire transfer", "needs_compliance_review": "needs compliance review"}, "vocabulary_predicates": ["has_compliance_clearance", "has_manual_exception", "may_receive_wire", "needs_compliance_review", "passed_kyc"], "query_atom": "may_receive_wire::Account 12", "target_statement": "Account 12 may receive a wire transfer.", "gold_reason_type": "entailed_by_fact_or_rule", "symbolic": {"positive_atoms": ["has_compliance_clearance::Account 12", "passed_kyc::Account 12"], "negative_atoms": [], "rules": [{"antecedents": ["passed_kyc::Account 12", "has_compliance_clearance::Account 12"], "conclusion": "may_receive_wire::Account 12", "text": "If an account passed KYC checks and has compliance clearance, then that account may receive a wire transfer."}], "query_atom": "may_receive_wire::Account 12", "closure_atoms": ["has_compliance_clearance::Account 12", "may_receive_wire::Account 12", "passed_kyc::Account 12"]}, "source_reports": {"agent_a": {"positive": ["Account 12 has compliance clearance."], "negative": []}, "agent_b": {"positive": ["Account 12 passed KYC checks."], "negative": []}}, "gold_truth_value": "true", "gold_source_used": "both", "gold_closure_handling": "entailed_or_explicit", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_compliance_clearance, may_receive_wire.\nAgent B open/incomplete predicates: has_manual_exception, needs_compliance_review, passed_kyc.\n\nPredicate glossary:\n- has_compliance_clearance: has compliance clearance\n- has_manual_exception: has a manual exception\n- may_receive_wire: may receive a wire transfer\n- needs_compliance_review: needs compliance review\n- passed_kyc: passed KYC checks\n\nCoordinator policy rules:\n- If an account passed KYC checks and has compliance clearance, then that account may receive a wire transfer.\n\nAgent A report:\nPositive facts:\n- Account 12 has compliance clearance.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Account 12 passed KYC checks.\nNegative facts:\n- none\n\nTarget atom: may_receive_wire::Account 12.\nTarget statement: Account 12 may receive a wire transfer.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "finance_controls_0144_entailed_closed_conclusion__owa__multi_agent", "source_id": "finance_controls_0144_entailed_closed_conclusion__owa", "base_id": "finance_controls_0144_entailed_closed_conclusion", "split": "train", "domain": "finance_controls", "family": "entailed_closed_conclusion", "subset": "control_entailed", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["has_compliance_clearance", "has_manual_exception", "may_receive_wire", "needs_compliance_review", "passed_kyc"], "facts_positive": ["Account 12 has compliance clearance.", "Account 12 passed KYC checks."], "facts_negative": [], "rules_natural": ["If an account passed KYC checks and has compliance clearance, then that account may receive a wire transfer."], "predicate_glossary": {"passed_kyc": "passed KYC checks", "has_compliance_clearance": "has compliance clearance", "has_manual_exception": "has a manual exception", "may_receive_wire": "may receive a wire transfer", "needs_compliance_review": "needs compliance review"}, "vocabulary_predicates": ["has_compliance_clearance", "has_manual_exception", "may_receive_wire", "needs_compliance_review", "passed_kyc"], "query_atom": "may_receive_wire::Account 12", "target_statement": "Account 12 may receive a wire transfer.", "gold_reason_type": "entailed_by_fact_or_rule", "symbolic": {"positive_atoms": ["has_compliance_clearance::Account 12", "passed_kyc::Account 12"], "negative_atoms": [], "rules": [{"antecedents": ["passed_kyc::Account 12", "has_compliance_clearance::Account 12"], "conclusion": "may_receive_wire::Account 12", "text": "If an account passed KYC checks and has compliance clearance, then that account may receive a wire transfer."}], "query_atom": "may_receive_wire::Account 12", "closure_atoms": ["has_compliance_clearance::Account 12", "may_receive_wire::Account 12", "passed_kyc::Account 12"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Account 12 has compliance clearance.", "Account 12 passed KYC checks."], "negative": []}}, "gold_truth_value": "true", "gold_source_used": "agent_b", "gold_closure_handling": "entailed_or_explicit", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: has_compliance_clearance, has_manual_exception, may_receive_wire, needs_compliance_review, passed_kyc.\n\nPredicate glossary:\n- has_compliance_clearance: has compliance clearance\n- has_manual_exception: has a manual exception\n- may_receive_wire: may receive a wire transfer\n- needs_compliance_review: needs compliance review\n- passed_kyc: passed KYC checks\n\nCoordinator policy rules:\n- If an account passed KYC checks and has compliance clearance, then that account may receive a wire transfer.\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Account 12 has compliance clearance.\n- Account 12 passed KYC checks.\nNegative facts:\n- none\n\nTarget atom: may_receive_wire::Account 12.\nTarget statement: Account 12 may receive a wire transfer.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "finance_controls_0146_entailed_closed_conclusion__cwa__multi_agent", "source_id": "finance_controls_0146_entailed_closed_conclusion__cwa", "base_id": "finance_controls_0146_entailed_closed_conclusion", "split": "test", "domain": "finance_controls", "family": "entailed_closed_conclusion", "subset": "control_entailed", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["has_compliance_clearance", "has_manual_exception", "may_receive_wire", "needs_compliance_review", "passed_kyc"], "agent_a_complete_predicates": ["has_compliance_clearance", "has_manual_exception", "may_receive_wire", "needs_compliance_review", "passed_kyc"], "agent_b_open_predicates": [], "facts_positive": ["Account 35 has compliance clearance.", "Account 35 passed KYC checks."], "facts_negative": [], "rules_natural": ["If an account passed KYC checks and has compliance clearance, then that account may receive a wire transfer."], "predicate_glossary": {"passed_kyc": "passed KYC checks", "has_compliance_clearance": "has compliance clearance", "has_manual_exception": "has a manual exception", "may_receive_wire": "may receive a wire transfer", "needs_compliance_review": "needs compliance review"}, "vocabulary_predicates": ["has_compliance_clearance", "has_manual_exception", "may_receive_wire", "needs_compliance_review", "passed_kyc"], "query_atom": "may_receive_wire::Account 35", "target_statement": "Account 35 may receive a wire transfer.", "gold_reason_type": "entailed_by_fact_or_rule", "symbolic": {"positive_atoms": ["has_compliance_clearance::Account 35", "passed_kyc::Account 35"], "negative_atoms": [], "rules": [{"antecedents": ["passed_kyc::Account 35", "has_compliance_clearance::Account 35"], "conclusion": "may_receive_wire::Account 35", "text": "If an account passed KYC checks and has compliance clearance, then that account may receive a wire transfer."}], "query_atom": "may_receive_wire::Account 35", "closure_atoms": ["has_compliance_clearance::Account 35", "may_receive_wire::Account 35", "passed_kyc::Account 35"]}, "source_reports": {"agent_a": {"positive": ["Account 35 has compliance clearance.", "Account 35 passed KYC checks."], "negative": []}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "true", "gold_source_used": "agent_a", "gold_closure_handling": "entailed_or_explicit", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_compliance_clearance, has_manual_exception, may_receive_wire, needs_compliance_review, passed_kyc.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- has_compliance_clearance: has compliance clearance\n- has_manual_exception: has a manual exception\n- may_receive_wire: may receive a wire transfer\n- needs_compliance_review: needs compliance review\n- passed_kyc: passed KYC checks\n\nCoordinator policy rules:\n- If an account passed KYC checks and has compliance clearance, then that account may receive a wire transfer.\n\nAgent A report:\nPositive facts:\n- Account 35 has compliance clearance.\n- Account 35 passed KYC checks.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: may_receive_wire::Account 35.\nTarget statement: Account 35 may receive a wire transfer.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "finance_controls_0146_entailed_closed_conclusion__lcwa__multi_agent", "source_id": "finance_controls_0146_entailed_closed_conclusion__lcwa", "base_id": "finance_controls_0146_entailed_closed_conclusion", "split": "test", "domain": "finance_controls", "family": "entailed_closed_conclusion", "subset": "control_entailed", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_compliance_clearance", "may_receive_wire"], "agent_a_complete_predicates": ["has_compliance_clearance", "may_receive_wire"], "agent_b_open_predicates": ["has_manual_exception", "needs_compliance_review", "passed_kyc"], "facts_positive": ["Account 35 has compliance clearance.", "Account 35 passed KYC checks."], "facts_negative": [], "rules_natural": ["If an account passed KYC checks and has compliance clearance, then that account may receive a wire transfer."], "predicate_glossary": {"passed_kyc": "passed KYC checks", "has_compliance_clearance": "has compliance clearance", "has_manual_exception": "has a manual exception", "may_receive_wire": "may receive a wire transfer", "needs_compliance_review": "needs compliance review"}, "vocabulary_predicates": ["has_compliance_clearance", "has_manual_exception", "may_receive_wire", "needs_compliance_review", "passed_kyc"], "query_atom": "may_receive_wire::Account 35", "target_statement": "Account 35 may receive a wire transfer.", "gold_reason_type": "entailed_by_fact_or_rule", "symbolic": {"positive_atoms": ["has_compliance_clearance::Account 35", "passed_kyc::Account 35"], "negative_atoms": [], "rules": [{"antecedents": ["passed_kyc::Account 35", "has_compliance_clearance::Account 35"], "conclusion": "may_receive_wire::Account 35", "text": "If an account passed KYC checks and has compliance clearance, then that account may receive a wire transfer."}], "query_atom": "may_receive_wire::Account 35", "closure_atoms": ["has_compliance_clearance::Account 35", "may_receive_wire::Account 35", "passed_kyc::Account 35"]}, "source_reports": {"agent_a": {"positive": ["Account 35 has compliance clearance."], "negative": []}, "agent_b": {"positive": ["Account 35 passed KYC checks."], "negative": []}}, "gold_truth_value": "true", "gold_source_used": "both", "gold_closure_handling": "entailed_or_explicit", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_compliance_clearance, may_receive_wire.\nAgent B open/incomplete predicates: has_manual_exception, needs_compliance_review, passed_kyc.\n\nPredicate glossary:\n- has_compliance_clearance: has compliance clearance\n- has_manual_exception: has a manual exception\n- may_receive_wire: may receive a wire transfer\n- needs_compliance_review: needs compliance review\n- passed_kyc: passed KYC checks\n\nCoordinator policy rules:\n- If an account passed KYC checks and has compliance clearance, then that account may receive a wire transfer.\n\nAgent A report:\nPositive facts:\n- Account 35 has compliance clearance.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Account 35 passed KYC checks.\nNegative facts:\n- none\n\nTarget atom: may_receive_wire::Account 35.\nTarget statement: Account 35 may receive a wire transfer.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "finance_controls_0146_entailed_closed_conclusion__owa__multi_agent", "source_id": "finance_controls_0146_entailed_closed_conclusion__owa", "base_id": "finance_controls_0146_entailed_closed_conclusion", "split": "test", "domain": "finance_controls", "family": "entailed_closed_conclusion", "subset": "control_entailed", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["has_compliance_clearance", "has_manual_exception", "may_receive_wire", "needs_compliance_review", "passed_kyc"], "facts_positive": ["Account 35 has compliance clearance.", "Account 35 passed KYC checks."], "facts_negative": [], "rules_natural": ["If an account passed KYC checks and has compliance clearance, then that account may receive a wire transfer."], "predicate_glossary": {"passed_kyc": "passed KYC checks", "has_compliance_clearance": "has compliance clearance", "has_manual_exception": "has a manual exception", "may_receive_wire": "may receive a wire transfer", "needs_compliance_review": "needs compliance review"}, "vocabulary_predicates": ["has_compliance_clearance", "has_manual_exception", "may_receive_wire", "needs_compliance_review", "passed_kyc"], "query_atom": "may_receive_wire::Account 35", "target_statement": "Account 35 may receive a wire transfer.", "gold_reason_type": "entailed_by_fact_or_rule", "symbolic": {"positive_atoms": ["has_compliance_clearance::Account 35", "passed_kyc::Account 35"], "negative_atoms": [], "rules": [{"antecedents": ["passed_kyc::Account 35", "has_compliance_clearance::Account 35"], "conclusion": "may_receive_wire::Account 35", "text": "If an account passed KYC checks and has compliance clearance, then that account may receive a wire transfer."}], "query_atom": "may_receive_wire::Account 35", "closure_atoms": ["has_compliance_clearance::Account 35", "may_receive_wire::Account 35", "passed_kyc::Account 35"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Account 35 has compliance clearance.", "Account 35 passed KYC checks."], "negative": []}}, "gold_truth_value": "true", "gold_source_used": "agent_b", "gold_closure_handling": "entailed_or_explicit", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: has_compliance_clearance, has_manual_exception, may_receive_wire, needs_compliance_review, passed_kyc.\n\nPredicate glossary:\n- has_compliance_clearance: has compliance clearance\n- has_manual_exception: has a manual exception\n- may_receive_wire: may receive a wire transfer\n- needs_compliance_review: needs compliance review\n- passed_kyc: passed KYC checks\n\nCoordinator policy rules:\n- If an account passed KYC checks and has compliance clearance, then that account may receive a wire transfer.\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Account 35 has compliance clearance.\n- Account 35 passed KYC checks.\nNegative facts:\n- none\n\nTarget atom: may_receive_wire::Account 35.\nTarget statement: Account 35 may receive a wire transfer.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "finance_controls_0156_explicit_negative_open__cwa__multi_agent", "source_id": "finance_controls_0156_explicit_negative_open__cwa", "base_id": "finance_controls_0156_explicit_negative_open", "split": "train", "domain": "finance_controls", "family": "explicit_negative_open", "subset": "control_explicit_negative", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["has_compliance_clearance", "has_manual_exception", "may_receive_wire", "needs_compliance_review", "passed_kyc"], "agent_a_complete_predicates": ["has_compliance_clearance", "has_manual_exception", "may_receive_wire", "needs_compliance_review", "passed_kyc"], "agent_b_open_predicates": [], "facts_positive": ["Account 35 passed KYC checks."], "facts_negative": ["Account 12 does not has a manual exception."], "rules_natural": [], "predicate_glossary": {"passed_kyc": "passed KYC checks", "has_compliance_clearance": "has compliance clearance", "has_manual_exception": "has a manual exception", "may_receive_wire": "may receive a wire transfer", "needs_compliance_review": "needs compliance review"}, "vocabulary_predicates": ["has_compliance_clearance", "has_manual_exception", "may_receive_wire", "needs_compliance_review", "passed_kyc"], "query_atom": "has_manual_exception::Account 12", "target_statement": "Account 12 has a manual exception.", "gold_reason_type": "explicit_negative_fact", "symbolic": {"positive_atoms": ["passed_kyc::Account 35"], "negative_atoms": ["has_manual_exception::Account 12"], "rules": [], "query_atom": "has_manual_exception::Account 12", "closure_atoms": ["passed_kyc::Account 35"]}, "source_reports": {"agent_a": {"positive": ["Account 35 passed KYC checks."], "negative": ["Account 12 does not has a manual exception."]}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "explicit_negative", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_compliance_clearance, has_manual_exception, may_receive_wire, needs_compliance_review, passed_kyc.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- has_compliance_clearance: has compliance clearance\n- has_manual_exception: has a manual exception\n- may_receive_wire: may receive a wire transfer\n- needs_compliance_review: needs compliance review\n- passed_kyc: passed KYC checks\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- Account 35 passed KYC checks.\nNegative facts:\n- Account 12 does not has a manual exception.\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: has_manual_exception::Account 12.\nTarget statement: Account 12 has a manual exception.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "finance_controls_0156_explicit_negative_open__lcwa__multi_agent", "source_id": "finance_controls_0156_explicit_negative_open__lcwa", "base_id": "finance_controls_0156_explicit_negative_open", "split": "train", "domain": "finance_controls", "family": "explicit_negative_open", "subset": "control_explicit_negative", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_compliance_clearance", "may_receive_wire"], "agent_a_complete_predicates": ["has_compliance_clearance", "may_receive_wire"], "agent_b_open_predicates": ["has_manual_exception", "needs_compliance_review", "passed_kyc"], "facts_positive": ["Account 35 passed KYC checks."], "facts_negative": ["Account 12 does not has a manual exception."], "rules_natural": [], "predicate_glossary": {"passed_kyc": "passed KYC checks", "has_compliance_clearance": "has compliance clearance", "has_manual_exception": "has a manual exception", "may_receive_wire": "may receive a wire transfer", "needs_compliance_review": "needs compliance review"}, "vocabulary_predicates": ["has_compliance_clearance", "has_manual_exception", "may_receive_wire", "needs_compliance_review", "passed_kyc"], "query_atom": "has_manual_exception::Account 12", "target_statement": "Account 12 has a manual exception.", "gold_reason_type": "explicit_negative_fact", "symbolic": {"positive_atoms": ["passed_kyc::Account 35"], "negative_atoms": ["has_manual_exception::Account 12"], "rules": [], "query_atom": "has_manual_exception::Account 12", "closure_atoms": ["passed_kyc::Account 35"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Account 35 passed KYC checks."], "negative": ["Account 12 does not has a manual exception."]}}, "gold_truth_value": "false", "gold_source_used": "agent_b", "gold_closure_handling": "explicit_negative", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_compliance_clearance, may_receive_wire.\nAgent B open/incomplete predicates: has_manual_exception, needs_compliance_review, passed_kyc.\n\nPredicate glossary:\n- has_compliance_clearance: has compliance clearance\n- has_manual_exception: has a manual exception\n- may_receive_wire: may receive a wire transfer\n- needs_compliance_review: needs compliance review\n- passed_kyc: passed KYC checks\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Account 35 passed KYC checks.\nNegative facts:\n- Account 12 does not has a manual exception.\n\nTarget atom: has_manual_exception::Account 12.\nTarget statement: Account 12 has a manual exception.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "finance_controls_0156_explicit_negative_open__owa__multi_agent", "source_id": "finance_controls_0156_explicit_negative_open__owa", "base_id": "finance_controls_0156_explicit_negative_open", "split": "train", "domain": "finance_controls", "family": "explicit_negative_open", "subset": "control_explicit_negative", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["has_compliance_clearance", "has_manual_exception", "may_receive_wire", "needs_compliance_review", "passed_kyc"], "facts_positive": ["Account 35 passed KYC checks."], "facts_negative": ["Account 12 does not has a manual exception."], "rules_natural": [], "predicate_glossary": {"passed_kyc": "passed KYC checks", "has_compliance_clearance": "has compliance clearance", "has_manual_exception": "has a manual exception", "may_receive_wire": "may receive a wire transfer", "needs_compliance_review": "needs compliance review"}, "vocabulary_predicates": ["has_compliance_clearance", "has_manual_exception", "may_receive_wire", "needs_compliance_review", "passed_kyc"], "query_atom": "has_manual_exception::Account 12", "target_statement": "Account 12 has a manual exception.", "gold_reason_type": "explicit_negative_fact", "symbolic": {"positive_atoms": ["passed_kyc::Account 35"], "negative_atoms": ["has_manual_exception::Account 12"], "rules": [], "query_atom": "has_manual_exception::Account 12", "closure_atoms": ["passed_kyc::Account 35"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Account 35 passed KYC checks."], "negative": ["Account 12 does not has a manual exception."]}}, "gold_truth_value": "false", "gold_source_used": "agent_b", "gold_closure_handling": "explicit_negative", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: has_compliance_clearance, has_manual_exception, may_receive_wire, needs_compliance_review, passed_kyc.\n\nPredicate glossary:\n- has_compliance_clearance: has compliance clearance\n- has_manual_exception: has a manual exception\n- may_receive_wire: may receive a wire transfer\n- needs_compliance_review: needs compliance review\n- passed_kyc: passed KYC checks\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Account 35 passed KYC checks.\nNegative facts:\n- Account 12 does not has a manual exception.\n\nTarget atom: has_manual_exception::Account 12.\nTarget statement: Account 12 has a manual exception.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "finance_controls_0158_explicit_negative_open__cwa__multi_agent", "source_id": "finance_controls_0158_explicit_negative_open__cwa", "base_id": "finance_controls_0158_explicit_negative_open", "split": "test", "domain": "finance_controls", "family": "explicit_negative_open", "subset": "control_explicit_negative", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["has_compliance_clearance", "has_manual_exception", "may_receive_wire", "needs_compliance_review", "passed_kyc"], "agent_a_complete_predicates": ["has_compliance_clearance", "has_manual_exception", "may_receive_wire", "needs_compliance_review", "passed_kyc"], "agent_b_open_predicates": [], "facts_positive": ["Account 59 passed KYC checks."], "facts_negative": ["Account 35 does not has a manual exception."], "rules_natural": [], "predicate_glossary": {"passed_kyc": "passed KYC checks", "has_compliance_clearance": "has compliance clearance", "has_manual_exception": "has a manual exception", "may_receive_wire": "may receive a wire transfer", "needs_compliance_review": "needs compliance review"}, "vocabulary_predicates": ["has_compliance_clearance", "has_manual_exception", "may_receive_wire", "needs_compliance_review", "passed_kyc"], "query_atom": "has_manual_exception::Account 35", "target_statement": "Account 35 has a manual exception.", "gold_reason_type": "explicit_negative_fact", "symbolic": {"positive_atoms": ["passed_kyc::Account 59"], "negative_atoms": ["has_manual_exception::Account 35"], "rules": [], "query_atom": "has_manual_exception::Account 35", "closure_atoms": ["passed_kyc::Account 59"]}, "source_reports": {"agent_a": {"positive": ["Account 59 passed KYC checks."], "negative": ["Account 35 does not has a manual exception."]}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "explicit_negative", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_compliance_clearance, has_manual_exception, may_receive_wire, needs_compliance_review, passed_kyc.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- has_compliance_clearance: has compliance clearance\n- has_manual_exception: has a manual exception\n- may_receive_wire: may receive a wire transfer\n- needs_compliance_review: needs compliance review\n- passed_kyc: passed KYC checks\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- Account 59 passed KYC checks.\nNegative facts:\n- Account 35 does not has a manual exception.\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: has_manual_exception::Account 35.\nTarget statement: Account 35 has a manual exception.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "finance_controls_0158_explicit_negative_open__lcwa__multi_agent", "source_id": "finance_controls_0158_explicit_negative_open__lcwa", "base_id": "finance_controls_0158_explicit_negative_open", "split": "test", "domain": "finance_controls", "family": "explicit_negative_open", "subset": "control_explicit_negative", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_compliance_clearance", "may_receive_wire"], "agent_a_complete_predicates": ["has_compliance_clearance", "may_receive_wire"], "agent_b_open_predicates": ["has_manual_exception", "needs_compliance_review", "passed_kyc"], "facts_positive": ["Account 59 passed KYC checks."], "facts_negative": ["Account 35 does not has a manual exception."], "rules_natural": [], "predicate_glossary": {"passed_kyc": "passed KYC checks", "has_compliance_clearance": "has compliance clearance", "has_manual_exception": "has a manual exception", "may_receive_wire": "may receive a wire transfer", "needs_compliance_review": "needs compliance review"}, "vocabulary_predicates": ["has_compliance_clearance", "has_manual_exception", "may_receive_wire", "needs_compliance_review", "passed_kyc"], "query_atom": "has_manual_exception::Account 35", "target_statement": "Account 35 has a manual exception.", "gold_reason_type": "explicit_negative_fact", "symbolic": {"positive_atoms": ["passed_kyc::Account 59"], "negative_atoms": ["has_manual_exception::Account 35"], "rules": [], "query_atom": "has_manual_exception::Account 35", "closure_atoms": ["passed_kyc::Account 59"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Account 59 passed KYC checks."], "negative": ["Account 35 does not has a manual exception."]}}, "gold_truth_value": "false", "gold_source_used": "agent_b", "gold_closure_handling": "explicit_negative", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_compliance_clearance, may_receive_wire.\nAgent B open/incomplete predicates: has_manual_exception, needs_compliance_review, passed_kyc.\n\nPredicate glossary:\n- has_compliance_clearance: has compliance clearance\n- has_manual_exception: has a manual exception\n- may_receive_wire: may receive a wire transfer\n- needs_compliance_review: needs compliance review\n- passed_kyc: passed KYC checks\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Account 59 passed KYC checks.\nNegative facts:\n- Account 35 does not has a manual exception.\n\nTarget atom: has_manual_exception::Account 35.\nTarget statement: Account 35 has a manual exception.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "finance_controls_0158_explicit_negative_open__owa__multi_agent", "source_id": "finance_controls_0158_explicit_negative_open__owa", "base_id": "finance_controls_0158_explicit_negative_open", "split": "test", "domain": "finance_controls", "family": "explicit_negative_open", "subset": "control_explicit_negative", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["has_compliance_clearance", "has_manual_exception", "may_receive_wire", "needs_compliance_review", "passed_kyc"], "facts_positive": ["Account 59 passed KYC checks."], "facts_negative": ["Account 35 does not has a manual exception."], "rules_natural": [], "predicate_glossary": {"passed_kyc": "passed KYC checks", "has_compliance_clearance": "has compliance clearance", "has_manual_exception": "has a manual exception", "may_receive_wire": "may receive a wire transfer", "needs_compliance_review": "needs compliance review"}, "vocabulary_predicates": ["has_compliance_clearance", "has_manual_exception", "may_receive_wire", "needs_compliance_review", "passed_kyc"], "query_atom": "has_manual_exception::Account 35", "target_statement": "Account 35 has a manual exception.", "gold_reason_type": "explicit_negative_fact", "symbolic": {"positive_atoms": ["passed_kyc::Account 59"], "negative_atoms": ["has_manual_exception::Account 35"], "rules": [], "query_atom": "has_manual_exception::Account 35", "closure_atoms": ["passed_kyc::Account 59"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Account 59 passed KYC checks."], "negative": ["Account 35 does not has a manual exception."]}}, "gold_truth_value": "false", "gold_source_used": "agent_b", "gold_closure_handling": "explicit_negative", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: has_compliance_clearance, has_manual_exception, may_receive_wire, needs_compliance_review, passed_kyc.\n\nPredicate glossary:\n- has_compliance_clearance: has compliance clearance\n- has_manual_exception: has a manual exception\n- may_receive_wire: may receive a wire transfer\n- needs_compliance_review: needs compliance review\n- passed_kyc: passed KYC checks\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Account 59 passed KYC checks.\nNegative facts:\n- Account 35 does not has a manual exception.\n\nTarget atom: has_manual_exception::Account 35.\nTarget statement: Account 35 has a manual exception.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "hospital_access_0001_closed_missing_direct__cwa__multi_agent", "source_id": "hospital_access_0001_closed_missing_direct__cwa", "base_id": "hospital_access_0001_closed_missing_direct", "split": "train", "domain": "hospital_access", "family": "closed_missing_direct", "subset": "core_contrastive", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["assigned_to_trial", "completed_training", "has_clearance", "may_enter_lab", "needs_supervisor_review"], "agent_a_complete_predicates": ["assigned_to_trial", "completed_training", "has_clearance", "may_enter_lab", "needs_supervisor_review"], "agent_b_open_predicates": [], "facts_positive": ["Ben has completed safety training."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"completed_training": "has completed safety training", "has_clearance": "has security clearance", "assigned_to_trial": "is assigned to the clinical trial", "may_enter_lab": "may enter Lab A", "needs_supervisor_review": "needs supervisor review"}, "vocabulary_predicates": ["assigned_to_trial", "completed_training", "has_clearance", "may_enter_lab", "needs_supervisor_review"], "query_atom": "has_clearance::Ben", "target_statement": "Ben has security clearance.", "gold_reason_type": "global_closed_world_underivable", "symbolic": {"positive_atoms": ["completed_training::Ben"], "negative_atoms": [], "rules": [], "query_atom": "has_clearance::Ben", "closure_atoms": ["completed_training::Ben"]}, "source_reports": {"agent_a": {"positive": ["Ben has completed safety training."], "negative": []}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "closed_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: assigned_to_trial, completed_training, has_clearance, may_enter_lab, needs_supervisor_review.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- assigned_to_trial: is assigned to the clinical trial\n- completed_training: has completed safety training\n- has_clearance: has security clearance\n- may_enter_lab: may enter Lab A\n- needs_supervisor_review: needs supervisor review\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- Ben has completed safety training.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: has_clearance::Ben.\nTarget statement: Ben has security clearance.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "hospital_access_0001_closed_missing_direct__lcwa__multi_agent", "source_id": "hospital_access_0001_closed_missing_direct__lcwa", "base_id": "hospital_access_0001_closed_missing_direct", "split": "train", "domain": "hospital_access", "family": "closed_missing_direct", "subset": "core_contrastive", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_clearance"], "agent_a_complete_predicates": ["has_clearance"], "agent_b_open_predicates": ["assigned_to_trial", "completed_training", "may_enter_lab", "needs_supervisor_review"], "facts_positive": ["Ben has completed safety training."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"completed_training": "has completed safety training", "has_clearance": "has security clearance", "assigned_to_trial": "is assigned to the clinical trial", "may_enter_lab": "may enter Lab A", "needs_supervisor_review": "needs supervisor review"}, "vocabulary_predicates": ["assigned_to_trial", "completed_training", "has_clearance", "may_enter_lab", "needs_supervisor_review"], "query_atom": "has_clearance::Ben", "target_statement": "Ben has security clearance.", "gold_reason_type": "local_closed_world_underivable", "symbolic": {"positive_atoms": ["completed_training::Ben"], "negative_atoms": [], "rules": [], "query_atom": "has_clearance::Ben", "closure_atoms": ["completed_training::Ben"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Ben has completed safety training."], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "closed_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_clearance.\nAgent B open/incomplete predicates: assigned_to_trial, completed_training, may_enter_lab, needs_supervisor_review.\n\nPredicate glossary:\n- assigned_to_trial: is assigned to the clinical trial\n- completed_training: has completed safety training\n- has_clearance: has security clearance\n- may_enter_lab: may enter Lab A\n- needs_supervisor_review: needs supervisor review\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Ben has completed safety training.\nNegative facts:\n- none\n\nTarget atom: has_clearance::Ben.\nTarget statement: Ben has security clearance.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "hospital_access_0001_closed_missing_direct__owa__multi_agent", "source_id": "hospital_access_0001_closed_missing_direct__owa", "base_id": "hospital_access_0001_closed_missing_direct", "split": "train", "domain": "hospital_access", "family": "closed_missing_direct", "subset": "core_contrastive", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["assigned_to_trial", "completed_training", "has_clearance", "may_enter_lab", "needs_supervisor_review"], "facts_positive": ["Ben has completed safety training."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"completed_training": "has completed safety training", "has_clearance": "has security clearance", "assigned_to_trial": "is assigned to the clinical trial", "may_enter_lab": "may enter Lab A", "needs_supervisor_review": "needs supervisor review"}, "vocabulary_predicates": ["assigned_to_trial", "completed_training", "has_clearance", "may_enter_lab", "needs_supervisor_review"], "query_atom": "has_clearance::Ben", "target_statement": "Ben has security clearance.", "gold_reason_type": "open_world_underivable", "symbolic": {"positive_atoms": ["completed_training::Ben"], "negative_atoms": [], "rules": [], "query_atom": "has_clearance::Ben", "closure_atoms": ["completed_training::Ben"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Ben has completed safety training."], "negative": []}}, "gold_truth_value": "unknown", "gold_source_used": "agent_b", "gold_closure_handling": "open_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: assigned_to_trial, completed_training, has_clearance, may_enter_lab, needs_supervisor_review.\n\nPredicate glossary:\n- assigned_to_trial: is assigned to the clinical trial\n- completed_training: has completed safety training\n- has_clearance: has security clearance\n- may_enter_lab: may enter Lab A\n- needs_supervisor_review: needs supervisor review\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Ben has completed safety training.\nNegative facts:\n- none\n\nTarget atom: has_clearance::Ben.\nTarget statement: Ben has security clearance.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "hospital_access_0003_closed_missing_direct__cwa__multi_agent", "source_id": "hospital_access_0003_closed_missing_direct__cwa", "base_id": "hospital_access_0003_closed_missing_direct", "split": "test", "domain": "hospital_access", "family": "closed_missing_direct", "subset": "core_contrastive", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["assigned_to_trial", "completed_training", "has_clearance", "may_enter_lab", "needs_supervisor_review"], "agent_a_complete_predicates": ["assigned_to_trial", "completed_training", "has_clearance", "may_enter_lab", "needs_supervisor_review"], "agent_b_open_predicates": [], "facts_positive": ["Dylan has completed safety training."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"completed_training": "has completed safety training", "has_clearance": "has security clearance", "assigned_to_trial": "is assigned to the clinical trial", "may_enter_lab": "may enter Lab A", "needs_supervisor_review": "needs supervisor review"}, "vocabulary_predicates": ["assigned_to_trial", "completed_training", "has_clearance", "may_enter_lab", "needs_supervisor_review"], "query_atom": "has_clearance::Dylan", "target_statement": "Dylan has security clearance.", "gold_reason_type": "global_closed_world_underivable", "symbolic": {"positive_atoms": ["completed_training::Dylan"], "negative_atoms": [], "rules": [], "query_atom": "has_clearance::Dylan", "closure_atoms": ["completed_training::Dylan"]}, "source_reports": {"agent_a": {"positive": ["Dylan has completed safety training."], "negative": []}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "closed_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: assigned_to_trial, completed_training, has_clearance, may_enter_lab, needs_supervisor_review.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- assigned_to_trial: is assigned to the clinical trial\n- completed_training: has completed safety training\n- has_clearance: has security clearance\n- may_enter_lab: may enter Lab A\n- needs_supervisor_review: needs supervisor review\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- Dylan has completed safety training.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: has_clearance::Dylan.\nTarget statement: Dylan has security clearance.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "hospital_access_0003_closed_missing_direct__lcwa__multi_agent", "source_id": "hospital_access_0003_closed_missing_direct__lcwa", "base_id": "hospital_access_0003_closed_missing_direct", "split": "test", "domain": "hospital_access", "family": "closed_missing_direct", "subset": "core_contrastive", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_clearance"], "agent_a_complete_predicates": ["has_clearance"], "agent_b_open_predicates": ["assigned_to_trial", "completed_training", "may_enter_lab", "needs_supervisor_review"], "facts_positive": ["Dylan has completed safety training."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"completed_training": "has completed safety training", "has_clearance": "has security clearance", "assigned_to_trial": "is assigned to the clinical trial", "may_enter_lab": "may enter Lab A", "needs_supervisor_review": "needs supervisor review"}, "vocabulary_predicates": ["assigned_to_trial", "completed_training", "has_clearance", "may_enter_lab", "needs_supervisor_review"], "query_atom": "has_clearance::Dylan", "target_statement": "Dylan has security clearance.", "gold_reason_type": "local_closed_world_underivable", "symbolic": {"positive_atoms": ["completed_training::Dylan"], "negative_atoms": [], "rules": [], "query_atom": "has_clearance::Dylan", "closure_atoms": ["completed_training::Dylan"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Dylan has completed safety training."], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "closed_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_clearance.\nAgent B open/incomplete predicates: assigned_to_trial, completed_training, may_enter_lab, needs_supervisor_review.\n\nPredicate glossary:\n- assigned_to_trial: is assigned to the clinical trial\n- completed_training: has completed safety training\n- has_clearance: has security clearance\n- may_enter_lab: may enter Lab A\n- needs_supervisor_review: needs supervisor review\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Dylan has completed safety training.\nNegative facts:\n- none\n\nTarget atom: has_clearance::Dylan.\nTarget statement: Dylan has security clearance.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "hospital_access_0003_closed_missing_direct__owa__multi_agent", "source_id": "hospital_access_0003_closed_missing_direct__owa", "base_id": "hospital_access_0003_closed_missing_direct", "split": "test", "domain": "hospital_access", "family": "closed_missing_direct", "subset": "core_contrastive", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["assigned_to_trial", "completed_training", "has_clearance", "may_enter_lab", "needs_supervisor_review"], "facts_positive": ["Dylan has completed safety training."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"completed_training": "has completed safety training", "has_clearance": "has security clearance", "assigned_to_trial": "is assigned to the clinical trial", "may_enter_lab": "may enter Lab A", "needs_supervisor_review": "needs supervisor review"}, "vocabulary_predicates": ["assigned_to_trial", "completed_training", "has_clearance", "may_enter_lab", "needs_supervisor_review"], "query_atom": "has_clearance::Dylan", "target_statement": "Dylan has security clearance.", "gold_reason_type": "open_world_underivable", "symbolic": {"positive_atoms": ["completed_training::Dylan"], "negative_atoms": [], "rules": [], "query_atom": "has_clearance::Dylan", "closure_atoms": ["completed_training::Dylan"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Dylan has completed safety training."], "negative": []}}, "gold_truth_value": "unknown", "gold_source_used": "agent_b", "gold_closure_handling": "open_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: assigned_to_trial, completed_training, has_clearance, may_enter_lab, needs_supervisor_review.\n\nPredicate glossary:\n- assigned_to_trial: is assigned to the clinical trial\n- completed_training: has completed safety training\n- has_clearance: has security clearance\n- may_enter_lab: may enter Lab A\n- needs_supervisor_review: needs supervisor review\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Dylan has completed safety training.\nNegative facts:\n- none\n\nTarget atom: has_clearance::Dylan.\nTarget statement: Dylan has security clearance.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "hospital_access_0005_open_missing_direct__cwa__multi_agent", "source_id": "hospital_access_0005_open_missing_direct__cwa", "base_id": "hospital_access_0005_open_missing_direct", "split": "train", "domain": "hospital_access", "family": "open_missing_direct", "subset": "core_contrastive", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["assigned_to_trial", "completed_training", "has_clearance", "may_enter_lab", "needs_supervisor_review"], "agent_a_complete_predicates": ["assigned_to_trial", "completed_training", "has_clearance", "may_enter_lab", "needs_supervisor_review"], "agent_b_open_predicates": [], "facts_positive": ["Farid has completed safety training."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"completed_training": "has completed safety training", "has_clearance": "has security clearance", "assigned_to_trial": "is assigned to the clinical trial", "may_enter_lab": "may enter Lab A", "needs_supervisor_review": "needs supervisor review"}, "vocabulary_predicates": ["assigned_to_trial", "completed_training", "has_clearance", "may_enter_lab", "needs_supervisor_review"], "query_atom": "assigned_to_trial::Farid", "target_statement": "Farid is assigned to the clinical trial.", "gold_reason_type": "global_closed_world_underivable", "symbolic": {"positive_atoms": ["completed_training::Farid"], "negative_atoms": [], "rules": [], "query_atom": "assigned_to_trial::Farid", "closure_atoms": ["completed_training::Farid"]}, "source_reports": {"agent_a": {"positive": ["Farid has completed safety training."], "negative": []}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "closed_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: assigned_to_trial, completed_training, has_clearance, may_enter_lab, needs_supervisor_review.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- assigned_to_trial: is assigned to the clinical trial\n- completed_training: has completed safety training\n- has_clearance: has security clearance\n- may_enter_lab: may enter Lab A\n- needs_supervisor_review: needs supervisor review\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- Farid has completed safety training.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: assigned_to_trial::Farid.\nTarget statement: Farid is assigned to the clinical trial.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "hospital_access_0005_open_missing_direct__lcwa__multi_agent", "source_id": "hospital_access_0005_open_missing_direct__lcwa", "base_id": "hospital_access_0005_open_missing_direct", "split": "train", "domain": "hospital_access", "family": "open_missing_direct", "subset": "core_contrastive", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_clearance", "may_enter_lab"], "agent_a_complete_predicates": ["has_clearance", "may_enter_lab"], "agent_b_open_predicates": ["assigned_to_trial", "completed_training", "needs_supervisor_review"], "facts_positive": ["Farid has completed safety training."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"completed_training": "has completed safety training", "has_clearance": "has security clearance", "assigned_to_trial": "is assigned to the clinical trial", "may_enter_lab": "may enter Lab A", "needs_supervisor_review": "needs supervisor review"}, "vocabulary_predicates": ["assigned_to_trial", "completed_training", "has_clearance", "may_enter_lab", "needs_supervisor_review"], "query_atom": "assigned_to_trial::Farid", "target_statement": "Farid is assigned to the clinical trial.", "gold_reason_type": "local_open_world_underivable", "symbolic": {"positive_atoms": ["completed_training::Farid"], "negative_atoms": [], "rules": [], "query_atom": "assigned_to_trial::Farid", "closure_atoms": ["completed_training::Farid"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Farid has completed safety training."], "negative": []}}, "gold_truth_value": "unknown", "gold_source_used": "agent_b", "gold_closure_handling": "open_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_clearance, may_enter_lab.\nAgent B open/incomplete predicates: assigned_to_trial, completed_training, needs_supervisor_review.\n\nPredicate glossary:\n- assigned_to_trial: is assigned to the clinical trial\n- completed_training: has completed safety training\n- has_clearance: has security clearance\n- may_enter_lab: may enter Lab A\n- needs_supervisor_review: needs supervisor review\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Farid has completed safety training.\nNegative facts:\n- none\n\nTarget atom: assigned_to_trial::Farid.\nTarget statement: Farid is assigned to the clinical trial.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "hospital_access_0005_open_missing_direct__owa__multi_agent", "source_id": "hospital_access_0005_open_missing_direct__owa", "base_id": "hospital_access_0005_open_missing_direct", "split": "train", "domain": "hospital_access", "family": "open_missing_direct", "subset": "core_contrastive", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["assigned_to_trial", "completed_training", "has_clearance", "may_enter_lab", "needs_supervisor_review"], "facts_positive": ["Farid has completed safety training."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"completed_training": "has completed safety training", "has_clearance": "has security clearance", "assigned_to_trial": "is assigned to the clinical trial", "may_enter_lab": "may enter Lab A", "needs_supervisor_review": "needs supervisor review"}, "vocabulary_predicates": ["assigned_to_trial", "completed_training", "has_clearance", "may_enter_lab", "needs_supervisor_review"], "query_atom": "assigned_to_trial::Farid", "target_statement": "Farid is assigned to the clinical trial.", "gold_reason_type": "open_world_underivable", "symbolic": {"positive_atoms": ["completed_training::Farid"], "negative_atoms": [], "rules": [], "query_atom": "assigned_to_trial::Farid", "closure_atoms": ["completed_training::Farid"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Farid has completed safety training."], "negative": []}}, "gold_truth_value": "unknown", "gold_source_used": "agent_b", "gold_closure_handling": "open_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: assigned_to_trial, completed_training, has_clearance, may_enter_lab, needs_supervisor_review.\n\nPredicate glossary:\n- assigned_to_trial: is assigned to the clinical trial\n- completed_training: has completed safety training\n- has_clearance: has security clearance\n- may_enter_lab: may enter Lab A\n- needs_supervisor_review: needs supervisor review\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Farid has completed safety training.\nNegative facts:\n- none\n\nTarget atom: assigned_to_trial::Farid.\nTarget statement: Farid is assigned to the clinical trial.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "hospital_access_0007_open_missing_direct__cwa__multi_agent", "source_id": "hospital_access_0007_open_missing_direct__cwa", "base_id": "hospital_access_0007_open_missing_direct", "split": "train", "domain": "hospital_access", "family": "open_missing_direct", "subset": "core_contrastive", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["assigned_to_trial", "completed_training", "has_clearance", "may_enter_lab", "needs_supervisor_review"], "agent_a_complete_predicates": ["assigned_to_trial", "completed_training", "has_clearance", "may_enter_lab", "needs_supervisor_review"], "agent_b_open_predicates": [], "facts_positive": ["Ben has completed safety training."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"completed_training": "has completed safety training", "has_clearance": "has security clearance", "assigned_to_trial": "is assigned to the clinical trial", "may_enter_lab": "may enter Lab A", "needs_supervisor_review": "needs supervisor review"}, "vocabulary_predicates": ["assigned_to_trial", "completed_training", "has_clearance", "may_enter_lab", "needs_supervisor_review"], "query_atom": "assigned_to_trial::Ben", "target_statement": "Ben is assigned to the clinical trial.", "gold_reason_type": "global_closed_world_underivable", "symbolic": {"positive_atoms": ["completed_training::Ben"], "negative_atoms": [], "rules": [], "query_atom": "assigned_to_trial::Ben", "closure_atoms": ["completed_training::Ben"]}, "source_reports": {"agent_a": {"positive": ["Ben has completed safety training."], "negative": []}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "closed_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: assigned_to_trial, completed_training, has_clearance, may_enter_lab, needs_supervisor_review.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- assigned_to_trial: is assigned to the clinical trial\n- completed_training: has completed safety training\n- has_clearance: has security clearance\n- may_enter_lab: may enter Lab A\n- needs_supervisor_review: needs supervisor review\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- Ben has completed safety training.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: assigned_to_trial::Ben.\nTarget statement: Ben is assigned to the clinical trial.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "hospital_access_0007_open_missing_direct__lcwa__multi_agent", "source_id": "hospital_access_0007_open_missing_direct__lcwa", "base_id": "hospital_access_0007_open_missing_direct", "split": "train", "domain": "hospital_access", "family": "open_missing_direct", "subset": "core_contrastive", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_clearance", "may_enter_lab"], "agent_a_complete_predicates": ["has_clearance", "may_enter_lab"], "agent_b_open_predicates": ["assigned_to_trial", "completed_training", "needs_supervisor_review"], "facts_positive": ["Ben has completed safety training."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"completed_training": "has completed safety training", "has_clearance": "has security clearance", "assigned_to_trial": "is assigned to the clinical trial", "may_enter_lab": "may enter Lab A", "needs_supervisor_review": "needs supervisor review"}, "vocabulary_predicates": ["assigned_to_trial", "completed_training", "has_clearance", "may_enter_lab", "needs_supervisor_review"], "query_atom": "assigned_to_trial::Ben", "target_statement": "Ben is assigned to the clinical trial.", "gold_reason_type": "local_open_world_underivable", "symbolic": {"positive_atoms": ["completed_training::Ben"], "negative_atoms": [], "rules": [], "query_atom": "assigned_to_trial::Ben", "closure_atoms": ["completed_training::Ben"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Ben has completed safety training."], "negative": []}}, "gold_truth_value": "unknown", "gold_source_used": "agent_b", "gold_closure_handling": "open_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_clearance, may_enter_lab.\nAgent B open/incomplete predicates: assigned_to_trial, completed_training, needs_supervisor_review.\n\nPredicate glossary:\n- assigned_to_trial: is assigned to the clinical trial\n- completed_training: has completed safety training\n- has_clearance: has security clearance\n- may_enter_lab: may enter Lab A\n- needs_supervisor_review: needs supervisor review\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Ben has completed safety training.\nNegative facts:\n- none\n\nTarget atom: assigned_to_trial::Ben.\nTarget statement: Ben is assigned to the clinical trial.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "hospital_access_0007_open_missing_direct__owa__multi_agent", "source_id": "hospital_access_0007_open_missing_direct__owa", "base_id": "hospital_access_0007_open_missing_direct", "split": "train", "domain": "hospital_access", "family": "open_missing_direct", "subset": "core_contrastive", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["assigned_to_trial", "completed_training", "has_clearance", "may_enter_lab", "needs_supervisor_review"], "facts_positive": ["Ben has completed safety training."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"completed_training": "has completed safety training", "has_clearance": "has security clearance", "assigned_to_trial": "is assigned to the clinical trial", "may_enter_lab": "may enter Lab A", "needs_supervisor_review": "needs supervisor review"}, "vocabulary_predicates": ["assigned_to_trial", "completed_training", "has_clearance", "may_enter_lab", "needs_supervisor_review"], "query_atom": "assigned_to_trial::Ben", "target_statement": "Ben is assigned to the clinical trial.", "gold_reason_type": "open_world_underivable", "symbolic": {"positive_atoms": ["completed_training::Ben"], "negative_atoms": [], "rules": [], "query_atom": "assigned_to_trial::Ben", "closure_atoms": ["completed_training::Ben"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Ben has completed safety training."], "negative": []}}, "gold_truth_value": "unknown", "gold_source_used": "agent_b", "gold_closure_handling": "open_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: assigned_to_trial, completed_training, has_clearance, may_enter_lab, needs_supervisor_review.\n\nPredicate glossary:\n- assigned_to_trial: is assigned to the clinical trial\n- completed_training: has completed safety training\n- has_clearance: has security clearance\n- may_enter_lab: may enter Lab A\n- needs_supervisor_review: needs supervisor review\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Ben has completed safety training.\nNegative facts:\n- none\n\nTarget atom: assigned_to_trial::Ben.\nTarget statement: Ben is assigned to the clinical trial.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "hospital_access_0012_open_missing_with_closed_distractor__cwa__multi_agent", "source_id": "hospital_access_0012_open_missing_with_closed_distractor__cwa", "base_id": "hospital_access_0012_open_missing_with_closed_distractor", "split": "train", "domain": "hospital_access", "family": "open_missing_with_closed_distractor", "subset": "core_contrastive", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["assigned_to_trial", "completed_training", "has_clearance", "may_enter_lab", "needs_supervisor_review"], "agent_a_complete_predicates": ["assigned_to_trial", "completed_training", "has_clearance", "may_enter_lab", "needs_supervisor_review"], "agent_b_open_predicates": [], "facts_positive": ["Ben has completed safety training.", "Ben has security clearance."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"completed_training": "has completed safety training", "has_clearance": "has security clearance", "assigned_to_trial": "is assigned to the clinical trial", "may_enter_lab": "may enter Lab A", "needs_supervisor_review": "needs supervisor review"}, "vocabulary_predicates": ["assigned_to_trial", "completed_training", "has_clearance", "may_enter_lab", "needs_supervisor_review"], "query_atom": "assigned_to_trial::Ava", "target_statement": "Ava is assigned to the clinical trial.", "gold_reason_type": "global_closed_world_underivable", "symbolic": {"positive_atoms": ["completed_training::Ben", "has_clearance::Ben"], "negative_atoms": [], "rules": [], "query_atom": "assigned_to_trial::Ava", "closure_atoms": ["completed_training::Ben", "has_clearance::Ben"]}, "source_reports": {"agent_a": {"positive": ["Ben has completed safety training.", "Ben has security clearance."], "negative": []}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "closed_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: assigned_to_trial, completed_training, has_clearance, may_enter_lab, needs_supervisor_review.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- assigned_to_trial: is assigned to the clinical trial\n- completed_training: has completed safety training\n- has_clearance: has security clearance\n- may_enter_lab: may enter Lab A\n- needs_supervisor_review: needs supervisor review\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- Ben has completed safety training.\n- Ben has security clearance.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: assigned_to_trial::Ava.\nTarget statement: Ava is assigned to the clinical trial.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "hospital_access_0012_open_missing_with_closed_distractor__lcwa__multi_agent", "source_id": "hospital_access_0012_open_missing_with_closed_distractor__lcwa", "base_id": "hospital_access_0012_open_missing_with_closed_distractor", "split": "train", "domain": "hospital_access", "family": "open_missing_with_closed_distractor", "subset": "core_contrastive", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_clearance", "may_enter_lab"], "agent_a_complete_predicates": ["has_clearance", "may_enter_lab"], "agent_b_open_predicates": ["assigned_to_trial", "completed_training", "needs_supervisor_review"], "facts_positive": ["Ben has completed safety training.", "Ben has security clearance."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"completed_training": "has completed safety training", "has_clearance": "has security clearance", "assigned_to_trial": "is assigned to the clinical trial", "may_enter_lab": "may enter Lab A", "needs_supervisor_review": "needs supervisor review"}, "vocabulary_predicates": ["assigned_to_trial", "completed_training", "has_clearance", "may_enter_lab", "needs_supervisor_review"], "query_atom": "assigned_to_trial::Ava", "target_statement": "Ava is assigned to the clinical trial.", "gold_reason_type": "local_open_world_underivable", "symbolic": {"positive_atoms": ["completed_training::Ben", "has_clearance::Ben"], "negative_atoms": [], "rules": [], "query_atom": "assigned_to_trial::Ava", "closure_atoms": ["completed_training::Ben", "has_clearance::Ben"]}, "source_reports": {"agent_a": {"positive": ["Ben has security clearance."], "negative": []}, "agent_b": {"positive": ["Ben has completed safety training."], "negative": []}}, "gold_truth_value": "unknown", "gold_source_used": "agent_b", "gold_closure_handling": "open_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_clearance, may_enter_lab.\nAgent B open/incomplete predicates: assigned_to_trial, completed_training, needs_supervisor_review.\n\nPredicate glossary:\n- assigned_to_trial: is assigned to the clinical trial\n- completed_training: has completed safety training\n- has_clearance: has security clearance\n- may_enter_lab: may enter Lab A\n- needs_supervisor_review: needs supervisor review\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- Ben has security clearance.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Ben has completed safety training.\nNegative facts:\n- none\n\nTarget atom: assigned_to_trial::Ava.\nTarget statement: Ava is assigned to the clinical trial.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "hospital_access_0012_open_missing_with_closed_distractor__owa__multi_agent", "source_id": "hospital_access_0012_open_missing_with_closed_distractor__owa", "base_id": "hospital_access_0012_open_missing_with_closed_distractor", "split": "train", "domain": "hospital_access", "family": "open_missing_with_closed_distractor", "subset": "core_contrastive", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["assigned_to_trial", "completed_training", "has_clearance", "may_enter_lab", "needs_supervisor_review"], "facts_positive": ["Ben has completed safety training.", "Ben has security clearance."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"completed_training": "has completed safety training", "has_clearance": "has security clearance", "assigned_to_trial": "is assigned to the clinical trial", "may_enter_lab": "may enter Lab A", "needs_supervisor_review": "needs supervisor review"}, "vocabulary_predicates": ["assigned_to_trial", "completed_training", "has_clearance", "may_enter_lab", "needs_supervisor_review"], "query_atom": "assigned_to_trial::Ava", "target_statement": "Ava is assigned to the clinical trial.", "gold_reason_type": "open_world_underivable", "symbolic": {"positive_atoms": ["completed_training::Ben", "has_clearance::Ben"], "negative_atoms": [], "rules": [], "query_atom": "assigned_to_trial::Ava", "closure_atoms": ["completed_training::Ben", "has_clearance::Ben"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Ben has completed safety training.", "Ben has security clearance."], "negative": []}}, "gold_truth_value": "unknown", "gold_source_used": "agent_b", "gold_closure_handling": "open_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: assigned_to_trial, completed_training, has_clearance, may_enter_lab, needs_supervisor_review.\n\nPredicate glossary:\n- assigned_to_trial: is assigned to the clinical trial\n- completed_training: has completed safety training\n- has_clearance: has security clearance\n- may_enter_lab: may enter Lab A\n- needs_supervisor_review: needs supervisor review\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Ben has completed safety training.\n- Ben has security clearance.\nNegative facts:\n- none\n\nTarget atom: assigned_to_trial::Ava.\nTarget statement: Ava is assigned to the clinical trial.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "hospital_access_0014_open_missing_with_closed_distractor__cwa__multi_agent", "source_id": "hospital_access_0014_open_missing_with_closed_distractor__cwa", "base_id": "hospital_access_0014_open_missing_with_closed_distractor", "split": "dev", "domain": "hospital_access", "family": "open_missing_with_closed_distractor", "subset": "core_contrastive", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["assigned_to_trial", "completed_training", "has_clearance", "may_enter_lab", "needs_supervisor_review"], "agent_a_complete_predicates": ["assigned_to_trial", "completed_training", "has_clearance", "may_enter_lab", "needs_supervisor_review"], "agent_b_open_predicates": [], "facts_positive": ["Dylan has completed safety training.", "Dylan has security clearance."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"completed_training": "has completed safety training", "has_clearance": "has security clearance", "assigned_to_trial": "is assigned to the clinical trial", "may_enter_lab": "may enter Lab A", "needs_supervisor_review": "needs supervisor review"}, "vocabulary_predicates": ["assigned_to_trial", "completed_training", "has_clearance", "may_enter_lab", "needs_supervisor_review"], "query_atom": "assigned_to_trial::Cora", "target_statement": "Cora is assigned to the clinical trial.", "gold_reason_type": "global_closed_world_underivable", "symbolic": {"positive_atoms": ["completed_training::Dylan", "has_clearance::Dylan"], "negative_atoms": [], "rules": [], "query_atom": "assigned_to_trial::Cora", "closure_atoms": ["completed_training::Dylan", "has_clearance::Dylan"]}, "source_reports": {"agent_a": {"positive": ["Dylan has completed safety training.", "Dylan has security clearance."], "negative": []}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "closed_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: assigned_to_trial, completed_training, has_clearance, may_enter_lab, needs_supervisor_review.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- assigned_to_trial: is assigned to the clinical trial\n- completed_training: has completed safety training\n- has_clearance: has security clearance\n- may_enter_lab: may enter Lab A\n- needs_supervisor_review: needs supervisor review\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- Dylan has completed safety training.\n- Dylan has security clearance.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: assigned_to_trial::Cora.\nTarget statement: Cora is assigned to the clinical trial.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "hospital_access_0014_open_missing_with_closed_distractor__lcwa__multi_agent", "source_id": "hospital_access_0014_open_missing_with_closed_distractor__lcwa", "base_id": "hospital_access_0014_open_missing_with_closed_distractor", "split": "dev", "domain": "hospital_access", "family": "open_missing_with_closed_distractor", "subset": "core_contrastive", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_clearance", "may_enter_lab"], "agent_a_complete_predicates": ["has_clearance", "may_enter_lab"], "agent_b_open_predicates": ["assigned_to_trial", "completed_training", "needs_supervisor_review"], "facts_positive": ["Dylan has completed safety training.", "Dylan has security clearance."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"completed_training": "has completed safety training", "has_clearance": "has security clearance", "assigned_to_trial": "is assigned to the clinical trial", "may_enter_lab": "may enter Lab A", "needs_supervisor_review": "needs supervisor review"}, "vocabulary_predicates": ["assigned_to_trial", "completed_training", "has_clearance", "may_enter_lab", "needs_supervisor_review"], "query_atom": "assigned_to_trial::Cora", "target_statement": "Cora is assigned to the clinical trial.", "gold_reason_type": "local_open_world_underivable", "symbolic": {"positive_atoms": ["completed_training::Dylan", "has_clearance::Dylan"], "negative_atoms": [], "rules": [], "query_atom": "assigned_to_trial::Cora", "closure_atoms": ["completed_training::Dylan", "has_clearance::Dylan"]}, "source_reports": {"agent_a": {"positive": ["Dylan has security clearance."], "negative": []}, "agent_b": {"positive": ["Dylan has completed safety training."], "negative": []}}, "gold_truth_value": "unknown", "gold_source_used": "agent_b", "gold_closure_handling": "open_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_clearance, may_enter_lab.\nAgent B open/incomplete predicates: assigned_to_trial, completed_training, needs_supervisor_review.\n\nPredicate glossary:\n- assigned_to_trial: is assigned to the clinical trial\n- completed_training: has completed safety training\n- has_clearance: has security clearance\n- may_enter_lab: may enter Lab A\n- needs_supervisor_review: needs supervisor review\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- Dylan has security clearance.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Dylan has completed safety training.\nNegative facts:\n- none\n\nTarget atom: assigned_to_trial::Cora.\nTarget statement: Cora is assigned to the clinical trial.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "hospital_access_0014_open_missing_with_closed_distractor__owa__multi_agent", "source_id": "hospital_access_0014_open_missing_with_closed_distractor__owa", "base_id": "hospital_access_0014_open_missing_with_closed_distractor", "split": "dev", "domain": "hospital_access", "family": "open_missing_with_closed_distractor", "subset": "core_contrastive", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["assigned_to_trial", "completed_training", "has_clearance", "may_enter_lab", "needs_supervisor_review"], "facts_positive": ["Dylan has completed safety training.", "Dylan has security clearance."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"completed_training": "has completed safety training", "has_clearance": "has security clearance", "assigned_to_trial": "is assigned to the clinical trial", "may_enter_lab": "may enter Lab A", "needs_supervisor_review": "needs supervisor review"}, "vocabulary_predicates": ["assigned_to_trial", "completed_training", "has_clearance", "may_enter_lab", "needs_supervisor_review"], "query_atom": "assigned_to_trial::Cora", "target_statement": "Cora is assigned to the clinical trial.", "gold_reason_type": "open_world_underivable", "symbolic": {"positive_atoms": ["completed_training::Dylan", "has_clearance::Dylan"], "negative_atoms": [], "rules": [], "query_atom": "assigned_to_trial::Cora", "closure_atoms": ["completed_training::Dylan", "has_clearance::Dylan"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Dylan has completed safety training.", "Dylan has security clearance."], "negative": []}}, "gold_truth_value": "unknown", "gold_source_used": "agent_b", "gold_closure_handling": "open_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: assigned_to_trial, completed_training, has_clearance, may_enter_lab, needs_supervisor_review.\n\nPredicate glossary:\n- assigned_to_trial: is assigned to the clinical trial\n- completed_training: has completed safety training\n- has_clearance: has security clearance\n- may_enter_lab: may enter Lab A\n- needs_supervisor_review: needs supervisor review\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Dylan has completed safety training.\n- Dylan has security clearance.\nNegative facts:\n- none\n\nTarget atom: assigned_to_trial::Cora.\nTarget statement: Cora is assigned to the clinical trial.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "hospital_access_0015_open_missing_with_closed_distractor__cwa__multi_agent", "source_id": "hospital_access_0015_open_missing_with_closed_distractor__cwa", "base_id": "hospital_access_0015_open_missing_with_closed_distractor", "split": "train", "domain": "hospital_access", "family": "open_missing_with_closed_distractor", "subset": "core_contrastive", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["assigned_to_trial", "completed_training", "has_clearance", "may_enter_lab", "needs_supervisor_review"], "agent_a_complete_predicates": ["assigned_to_trial", "completed_training", "has_clearance", "may_enter_lab", "needs_supervisor_review"], "agent_b_open_predicates": [], "facts_positive": ["Elena has completed safety training.", "Elena has security clearance."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"completed_training": "has completed safety training", "has_clearance": "has security clearance", "assigned_to_trial": "is assigned to the clinical trial", "may_enter_lab": "may enter Lab A", "needs_supervisor_review": "needs supervisor review"}, "vocabulary_predicates": ["assigned_to_trial", "completed_training", "has_clearance", "may_enter_lab", "needs_supervisor_review"], "query_atom": "assigned_to_trial::Dylan", "target_statement": "Dylan is assigned to the clinical trial.", "gold_reason_type": "global_closed_world_underivable", "symbolic": {"positive_atoms": ["completed_training::Elena", "has_clearance::Elena"], "negative_atoms": [], "rules": [], "query_atom": "assigned_to_trial::Dylan", "closure_atoms": ["completed_training::Elena", "has_clearance::Elena"]}, "source_reports": {"agent_a": {"positive": ["Elena has completed safety training.", "Elena has security clearance."], "negative": []}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "closed_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: assigned_to_trial, completed_training, has_clearance, may_enter_lab, needs_supervisor_review.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- assigned_to_trial: is assigned to the clinical trial\n- completed_training: has completed safety training\n- has_clearance: has security clearance\n- may_enter_lab: may enter Lab A\n- needs_supervisor_review: needs supervisor review\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- Elena has completed safety training.\n- Elena has security clearance.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: assigned_to_trial::Dylan.\nTarget statement: Dylan is assigned to the clinical trial.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "hospital_access_0015_open_missing_with_closed_distractor__lcwa__multi_agent", "source_id": "hospital_access_0015_open_missing_with_closed_distractor__lcwa", "base_id": "hospital_access_0015_open_missing_with_closed_distractor", "split": "train", "domain": "hospital_access", "family": "open_missing_with_closed_distractor", "subset": "core_contrastive", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_clearance", "may_enter_lab"], "agent_a_complete_predicates": ["has_clearance", "may_enter_lab"], "agent_b_open_predicates": ["assigned_to_trial", "completed_training", "needs_supervisor_review"], "facts_positive": ["Elena has completed safety training.", "Elena has security clearance."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"completed_training": "has completed safety training", "has_clearance": "has security clearance", "assigned_to_trial": "is assigned to the clinical trial", "may_enter_lab": "may enter Lab A", "needs_supervisor_review": "needs supervisor review"}, "vocabulary_predicates": ["assigned_to_trial", "completed_training", "has_clearance", "may_enter_lab", "needs_supervisor_review"], "query_atom": "assigned_to_trial::Dylan", "target_statement": "Dylan is assigned to the clinical trial.", "gold_reason_type": "local_open_world_underivable", "symbolic": {"positive_atoms": ["completed_training::Elena", "has_clearance::Elena"], "negative_atoms": [], "rules": [], "query_atom": "assigned_to_trial::Dylan", "closure_atoms": ["completed_training::Elena", "has_clearance::Elena"]}, "source_reports": {"agent_a": {"positive": ["Elena has security clearance."], "negative": []}, "agent_b": {"positive": ["Elena has completed safety training."], "negative": []}}, "gold_truth_value": "unknown", "gold_source_used": "agent_b", "gold_closure_handling": "open_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_clearance, may_enter_lab.\nAgent B open/incomplete predicates: assigned_to_trial, completed_training, needs_supervisor_review.\n\nPredicate glossary:\n- assigned_to_trial: is assigned to the clinical trial\n- completed_training: has completed safety training\n- has_clearance: has security clearance\n- may_enter_lab: may enter Lab A\n- needs_supervisor_review: needs supervisor review\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- Elena has security clearance.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Elena has completed safety training.\nNegative facts:\n- none\n\nTarget atom: assigned_to_trial::Dylan.\nTarget statement: Dylan is assigned to the clinical trial.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "hospital_access_0015_open_missing_with_closed_distractor__owa__multi_agent", "source_id": "hospital_access_0015_open_missing_with_closed_distractor__owa", "base_id": "hospital_access_0015_open_missing_with_closed_distractor", "split": "train", "domain": "hospital_access", "family": "open_missing_with_closed_distractor", "subset": "core_contrastive", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["assigned_to_trial", "completed_training", "has_clearance", "may_enter_lab", "needs_supervisor_review"], "facts_positive": ["Elena has completed safety training.", "Elena has security clearance."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"completed_training": "has completed safety training", "has_clearance": "has security clearance", "assigned_to_trial": "is assigned to the clinical trial", "may_enter_lab": "may enter Lab A", "needs_supervisor_review": "needs supervisor review"}, "vocabulary_predicates": ["assigned_to_trial", "completed_training", "has_clearance", "may_enter_lab", "needs_supervisor_review"], "query_atom": "assigned_to_trial::Dylan", "target_statement": "Dylan is assigned to the clinical trial.", "gold_reason_type": "open_world_underivable", "symbolic": {"positive_atoms": ["completed_training::Elena", "has_clearance::Elena"], "negative_atoms": [], "rules": [], "query_atom": "assigned_to_trial::Dylan", "closure_atoms": ["completed_training::Elena", "has_clearance::Elena"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Elena has completed safety training.", "Elena has security clearance."], "negative": []}}, "gold_truth_value": "unknown", "gold_source_used": "agent_b", "gold_closure_handling": "open_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: assigned_to_trial, completed_training, has_clearance, may_enter_lab, needs_supervisor_review.\n\nPredicate glossary:\n- assigned_to_trial: is assigned to the clinical trial\n- completed_training: has completed safety training\n- has_clearance: has security clearance\n- may_enter_lab: may enter Lab A\n- needs_supervisor_review: needs supervisor review\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Elena has completed safety training.\n- Elena has security clearance.\nNegative facts:\n- none\n\nTarget atom: assigned_to_trial::Dylan.\nTarget statement: Dylan is assigned to the clinical trial.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "hospital_access_0022_open_derived_missing_antecedent__cwa__multi_agent", "source_id": "hospital_access_0022_open_derived_missing_antecedent__cwa", "base_id": "hospital_access_0022_open_derived_missing_antecedent", "split": "train", "domain": "hospital_access", "family": "open_derived_missing_antecedent", "subset": "core_contrastive", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["assigned_to_trial", "completed_training", "has_clearance", "may_enter_lab", "needs_supervisor_review"], "agent_a_complete_predicates": ["assigned_to_trial", "completed_training", "has_clearance", "may_enter_lab", "needs_supervisor_review"], "agent_b_open_predicates": [], "facts_positive": ["Elena has completed safety training."], "facts_negative": [], "rules_natural": ["If a staff member is assigned to the clinical trial, then that staff member needs supervisor review."], "predicate_glossary": {"completed_training": "has completed safety training", "has_clearance": "has security clearance", "assigned_to_trial": "is assigned to the clinical trial", "may_enter_lab": "may enter Lab A", "needs_supervisor_review": "needs supervisor review"}, "vocabulary_predicates": ["assigned_to_trial", "completed_training", "has_clearance", "may_enter_lab", "needs_supervisor_review"], "query_atom": "needs_supervisor_review::Elena", "target_statement": "Elena needs supervisor review.", "gold_reason_type": "global_closed_world_underivable", "symbolic": {"positive_atoms": ["completed_training::Elena"], "negative_atoms": [], "rules": [{"antecedents": ["assigned_to_trial::Elena"], "conclusion": "needs_supervisor_review::Elena", "text": "If a staff member is assigned to the clinical trial, then that staff member needs supervisor review."}], "query_atom": "needs_supervisor_review::Elena", "closure_atoms": ["completed_training::Elena"]}, "source_reports": {"agent_a": {"positive": ["Elena has completed safety training."], "negative": []}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "closed_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: assigned_to_trial, completed_training, has_clearance, may_enter_lab, needs_supervisor_review.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- assigned_to_trial: is assigned to the clinical trial\n- completed_training: has completed safety training\n- has_clearance: has security clearance\n- may_enter_lab: may enter Lab A\n- needs_supervisor_review: needs supervisor review\n\nCoordinator policy rules:\n- If a staff member is assigned to the clinical trial, then that staff member needs supervisor review.\n\nAgent A report:\nPositive facts:\n- Elena has completed safety training.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: needs_supervisor_review::Elena.\nTarget statement: Elena needs supervisor review.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "hospital_access_0022_open_derived_missing_antecedent__lcwa__multi_agent", "source_id": "hospital_access_0022_open_derived_missing_antecedent__lcwa", "base_id": "hospital_access_0022_open_derived_missing_antecedent", "split": "train", "domain": "hospital_access", "family": "open_derived_missing_antecedent", "subset": "core_contrastive", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_clearance", "may_enter_lab"], "agent_a_complete_predicates": ["has_clearance", "may_enter_lab"], "agent_b_open_predicates": ["assigned_to_trial", "completed_training", "needs_supervisor_review"], "facts_positive": ["Elena has completed safety training."], "facts_negative": [], "rules_natural": ["If a staff member is assigned to the clinical trial, then that staff member needs supervisor review."], "predicate_glossary": {"completed_training": "has completed safety training", "has_clearance": "has security clearance", "assigned_to_trial": "is assigned to the clinical trial", "may_enter_lab": "may enter Lab A", "needs_supervisor_review": "needs supervisor review"}, "vocabulary_predicates": ["assigned_to_trial", "completed_training", "has_clearance", "may_enter_lab", "needs_supervisor_review"], "query_atom": "needs_supervisor_review::Elena", "target_statement": "Elena needs supervisor review.", "gold_reason_type": "local_open_world_underivable", "symbolic": {"positive_atoms": ["completed_training::Elena"], "negative_atoms": [], "rules": [{"antecedents": ["assigned_to_trial::Elena"], "conclusion": "needs_supervisor_review::Elena", "text": "If a staff member is assigned to the clinical trial, then that staff member needs supervisor review."}], "query_atom": "needs_supervisor_review::Elena", "closure_atoms": ["completed_training::Elena"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Elena has completed safety training."], "negative": []}}, "gold_truth_value": "unknown", "gold_source_used": "agent_b", "gold_closure_handling": "open_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_clearance, may_enter_lab.\nAgent B open/incomplete predicates: assigned_to_trial, completed_training, needs_supervisor_review.\n\nPredicate glossary:\n- assigned_to_trial: is assigned to the clinical trial\n- completed_training: has completed safety training\n- has_clearance: has security clearance\n- may_enter_lab: may enter Lab A\n- needs_supervisor_review: needs supervisor review\n\nCoordinator policy rules:\n- If a staff member is assigned to the clinical trial, then that staff member needs supervisor review.\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Elena has completed safety training.\nNegative facts:\n- none\n\nTarget atom: needs_supervisor_review::Elena.\nTarget statement: Elena needs supervisor review.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "hospital_access_0022_open_derived_missing_antecedent__owa__multi_agent", "source_id": "hospital_access_0022_open_derived_missing_antecedent__owa", "base_id": "hospital_access_0022_open_derived_missing_antecedent", "split": "train", "domain": "hospital_access", "family": "open_derived_missing_antecedent", "subset": "core_contrastive", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["assigned_to_trial", "completed_training", "has_clearance", "may_enter_lab", "needs_supervisor_review"], "facts_positive": ["Elena has completed safety training."], "facts_negative": [], "rules_natural": ["If a staff member is assigned to the clinical trial, then that staff member needs supervisor review."], "predicate_glossary": {"completed_training": "has completed safety training", "has_clearance": "has security clearance", "assigned_to_trial": "is assigned to the clinical trial", "may_enter_lab": "may enter Lab A", "needs_supervisor_review": "needs supervisor review"}, "vocabulary_predicates": ["assigned_to_trial", "completed_training", "has_clearance", "may_enter_lab", "needs_supervisor_review"], "query_atom": "needs_supervisor_review::Elena", "target_statement": "Elena needs supervisor review.", "gold_reason_type": "open_world_underivable", "symbolic": {"positive_atoms": ["completed_training::Elena"], "negative_atoms": [], "rules": [{"antecedents": ["assigned_to_trial::Elena"], "conclusion": "needs_supervisor_review::Elena", "text": "If a staff member is assigned to the clinical trial, then that staff member needs supervisor review."}], "query_atom": "needs_supervisor_review::Elena", "closure_atoms": ["completed_training::Elena"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Elena has completed safety training."], "negative": []}}, "gold_truth_value": "unknown", "gold_source_used": "agent_b", "gold_closure_handling": "open_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: assigned_to_trial, completed_training, has_clearance, may_enter_lab, needs_supervisor_review.\n\nPredicate glossary:\n- assigned_to_trial: is assigned to the clinical trial\n- completed_training: has completed safety training\n- has_clearance: has security clearance\n- may_enter_lab: may enter Lab A\n- needs_supervisor_review: needs supervisor review\n\nCoordinator policy rules:\n- If a staff member is assigned to the clinical trial, then that staff member needs supervisor review.\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Elena has completed safety training.\nNegative facts:\n- none\n\nTarget atom: needs_supervisor_review::Elena.\nTarget statement: Elena needs supervisor review.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "hospital_access_0023_open_derived_missing_antecedent__cwa__multi_agent", "source_id": "hospital_access_0023_open_derived_missing_antecedent__cwa", "base_id": "hospital_access_0023_open_derived_missing_antecedent", "split": "train", "domain": "hospital_access", "family": "open_derived_missing_antecedent", "subset": "core_contrastive", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["assigned_to_trial", "completed_training", "has_clearance", "may_enter_lab", "needs_supervisor_review"], "agent_a_complete_predicates": ["assigned_to_trial", "completed_training", "has_clearance", "may_enter_lab", "needs_supervisor_review"], "agent_b_open_predicates": [], "facts_positive": ["Farid has completed safety training."], "facts_negative": [], "rules_natural": ["If a staff member is assigned to the clinical trial, then that staff member needs supervisor review."], "predicate_glossary": {"completed_training": "has completed safety training", "has_clearance": "has security clearance", "assigned_to_trial": "is assigned to the clinical trial", "may_enter_lab": "may enter Lab A", "needs_supervisor_review": "needs supervisor review"}, "vocabulary_predicates": ["assigned_to_trial", "completed_training", "has_clearance", "may_enter_lab", "needs_supervisor_review"], "query_atom": "needs_supervisor_review::Farid", "target_statement": "Farid needs supervisor review.", "gold_reason_type": "global_closed_world_underivable", "symbolic": {"positive_atoms": ["completed_training::Farid"], "negative_atoms": [], "rules": [{"antecedents": ["assigned_to_trial::Farid"], "conclusion": "needs_supervisor_review::Farid", "text": "If a staff member is assigned to the clinical trial, then that staff member needs supervisor review."}], "query_atom": "needs_supervisor_review::Farid", "closure_atoms": ["completed_training::Farid"]}, "source_reports": {"agent_a": {"positive": ["Farid has completed safety training."], "negative": []}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "closed_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: assigned_to_trial, completed_training, has_clearance, may_enter_lab, needs_supervisor_review.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- assigned_to_trial: is assigned to the clinical trial\n- completed_training: has completed safety training\n- has_clearance: has security clearance\n- may_enter_lab: may enter Lab A\n- needs_supervisor_review: needs supervisor review\n\nCoordinator policy rules:\n- If a staff member is assigned to the clinical trial, then that staff member needs supervisor review.\n\nAgent A report:\nPositive facts:\n- Farid has completed safety training.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: needs_supervisor_review::Farid.\nTarget statement: Farid needs supervisor review.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "hospital_access_0023_open_derived_missing_antecedent__lcwa__multi_agent", "source_id": "hospital_access_0023_open_derived_missing_antecedent__lcwa", "base_id": "hospital_access_0023_open_derived_missing_antecedent", "split": "train", "domain": "hospital_access", "family": "open_derived_missing_antecedent", "subset": "core_contrastive", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_clearance", "may_enter_lab"], "agent_a_complete_predicates": ["has_clearance", "may_enter_lab"], "agent_b_open_predicates": ["assigned_to_trial", "completed_training", "needs_supervisor_review"], "facts_positive": ["Farid has completed safety training."], "facts_negative": [], "rules_natural": ["If a staff member is assigned to the clinical trial, then that staff member needs supervisor review."], "predicate_glossary": {"completed_training": "has completed safety training", "has_clearance": "has security clearance", "assigned_to_trial": "is assigned to the clinical trial", "may_enter_lab": "may enter Lab A", "needs_supervisor_review": "needs supervisor review"}, "vocabulary_predicates": ["assigned_to_trial", "completed_training", "has_clearance", "may_enter_lab", "needs_supervisor_review"], "query_atom": "needs_supervisor_review::Farid", "target_statement": "Farid needs supervisor review.", "gold_reason_type": "local_open_world_underivable", "symbolic": {"positive_atoms": ["completed_training::Farid"], "negative_atoms": [], "rules": [{"antecedents": ["assigned_to_trial::Farid"], "conclusion": "needs_supervisor_review::Farid", "text": "If a staff member is assigned to the clinical trial, then that staff member needs supervisor review."}], "query_atom": "needs_supervisor_review::Farid", "closure_atoms": ["completed_training::Farid"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Farid has completed safety training."], "negative": []}}, "gold_truth_value": "unknown", "gold_source_used": "agent_b", "gold_closure_handling": "open_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_clearance, may_enter_lab.\nAgent B open/incomplete predicates: assigned_to_trial, completed_training, needs_supervisor_review.\n\nPredicate glossary:\n- assigned_to_trial: is assigned to the clinical trial\n- completed_training: has completed safety training\n- has_clearance: has security clearance\n- may_enter_lab: may enter Lab A\n- needs_supervisor_review: needs supervisor review\n\nCoordinator policy rules:\n- If a staff member is assigned to the clinical trial, then that staff member needs supervisor review.\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Farid has completed safety training.\nNegative facts:\n- none\n\nTarget atom: needs_supervisor_review::Farid.\nTarget statement: Farid needs supervisor review.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "hospital_access_0023_open_derived_missing_antecedent__owa__multi_agent", "source_id": "hospital_access_0023_open_derived_missing_antecedent__owa", "base_id": "hospital_access_0023_open_derived_missing_antecedent", "split": "train", "domain": "hospital_access", "family": "open_derived_missing_antecedent", "subset": "core_contrastive", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["assigned_to_trial", "completed_training", "has_clearance", "may_enter_lab", "needs_supervisor_review"], "facts_positive": ["Farid has completed safety training."], "facts_negative": [], "rules_natural": ["If a staff member is assigned to the clinical trial, then that staff member needs supervisor review."], "predicate_glossary": {"completed_training": "has completed safety training", "has_clearance": "has security clearance", "assigned_to_trial": "is assigned to the clinical trial", "may_enter_lab": "may enter Lab A", "needs_supervisor_review": "needs supervisor review"}, "vocabulary_predicates": ["assigned_to_trial", "completed_training", "has_clearance", "may_enter_lab", "needs_supervisor_review"], "query_atom": "needs_supervisor_review::Farid", "target_statement": "Farid needs supervisor review.", "gold_reason_type": "open_world_underivable", "symbolic": {"positive_atoms": ["completed_training::Farid"], "negative_atoms": [], "rules": [{"antecedents": ["assigned_to_trial::Farid"], "conclusion": "needs_supervisor_review::Farid", "text": "If a staff member is assigned to the clinical trial, then that staff member needs supervisor review."}], "query_atom": "needs_supervisor_review::Farid", "closure_atoms": ["completed_training::Farid"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Farid has completed safety training."], "negative": []}}, "gold_truth_value": "unknown", "gold_source_used": "agent_b", "gold_closure_handling": "open_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: assigned_to_trial, completed_training, has_clearance, may_enter_lab, needs_supervisor_review.\n\nPredicate glossary:\n- assigned_to_trial: is assigned to the clinical trial\n- completed_training: has completed safety training\n- has_clearance: has security clearance\n- may_enter_lab: may enter Lab A\n- needs_supervisor_review: needs supervisor review\n\nCoordinator policy rules:\n- If a staff member is assigned to the clinical trial, then that staff member needs supervisor review.\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Farid has completed safety training.\nNegative facts:\n- none\n\nTarget atom: needs_supervisor_review::Farid.\nTarget statement: Farid needs supervisor review.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "hospital_access_0027_entailed_closed_conclusion__cwa__multi_agent", "source_id": "hospital_access_0027_entailed_closed_conclusion__cwa", "base_id": "hospital_access_0027_entailed_closed_conclusion", "split": "train", "domain": "hospital_access", "family": "entailed_closed_conclusion", "subset": "control_entailed", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["assigned_to_trial", "completed_training", "has_clearance", "may_enter_lab", "needs_supervisor_review"], "agent_a_complete_predicates": ["assigned_to_trial", "completed_training", "has_clearance", "may_enter_lab", "needs_supervisor_review"], "agent_b_open_predicates": [], "facts_positive": ["Dylan has completed safety training.", "Dylan has security clearance."], "facts_negative": [], "rules_natural": ["If a staff member has security clearance and has completed safety training, then that staff member may enter Lab A."], "predicate_glossary": {"completed_training": "has completed safety training", "has_clearance": "has security clearance", "assigned_to_trial": "is assigned to the clinical trial", "may_enter_lab": "may enter Lab A", "needs_supervisor_review": "needs supervisor review"}, "vocabulary_predicates": ["assigned_to_trial", "completed_training", "has_clearance", "may_enter_lab", "needs_supervisor_review"], "query_atom": "may_enter_lab::Dylan", "target_statement": "Dylan may enter Lab A.", "gold_reason_type": "entailed_by_fact_or_rule", "symbolic": {"positive_atoms": ["completed_training::Dylan", "has_clearance::Dylan"], "negative_atoms": [], "rules": [{"antecedents": ["completed_training::Dylan", "has_clearance::Dylan"], "conclusion": "may_enter_lab::Dylan", "text": "If a staff member has security clearance and has completed safety training, then that staff member may enter Lab A."}], "query_atom": "may_enter_lab::Dylan", "closure_atoms": ["completed_training::Dylan", "has_clearance::Dylan", "may_enter_lab::Dylan"]}, "source_reports": {"agent_a": {"positive": ["Dylan has completed safety training.", "Dylan has security clearance."], "negative": []}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "true", "gold_source_used": "agent_a", "gold_closure_handling": "entailed_or_explicit", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: assigned_to_trial, completed_training, has_clearance, may_enter_lab, needs_supervisor_review.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- assigned_to_trial: is assigned to the clinical trial\n- completed_training: has completed safety training\n- has_clearance: has security clearance\n- may_enter_lab: may enter Lab A\n- needs_supervisor_review: needs supervisor review\n\nCoordinator policy rules:\n- If a staff member has security clearance and has completed safety training, then that staff member may enter Lab A.\n\nAgent A report:\nPositive facts:\n- Dylan has completed safety training.\n- Dylan has security clearance.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: may_enter_lab::Dylan.\nTarget statement: Dylan may enter Lab A.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "hospital_access_0027_entailed_closed_conclusion__lcwa__multi_agent", "source_id": "hospital_access_0027_entailed_closed_conclusion__lcwa", "base_id": "hospital_access_0027_entailed_closed_conclusion", "split": "train", "domain": "hospital_access", "family": "entailed_closed_conclusion", "subset": "control_entailed", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_clearance", "may_enter_lab"], "agent_a_complete_predicates": ["has_clearance", "may_enter_lab"], "agent_b_open_predicates": ["assigned_to_trial", "completed_training", "needs_supervisor_review"], "facts_positive": ["Dylan has completed safety training.", "Dylan has security clearance."], "facts_negative": [], "rules_natural": ["If a staff member has security clearance and has completed safety training, then that staff member may enter Lab A."], "predicate_glossary": {"completed_training": "has completed safety training", "has_clearance": "has security clearance", "assigned_to_trial": "is assigned to the clinical trial", "may_enter_lab": "may enter Lab A", "needs_supervisor_review": "needs supervisor review"}, "vocabulary_predicates": ["assigned_to_trial", "completed_training", "has_clearance", "may_enter_lab", "needs_supervisor_review"], "query_atom": "may_enter_lab::Dylan", "target_statement": "Dylan may enter Lab A.", "gold_reason_type": "entailed_by_fact_or_rule", "symbolic": {"positive_atoms": ["completed_training::Dylan", "has_clearance::Dylan"], "negative_atoms": [], "rules": [{"antecedents": ["completed_training::Dylan", "has_clearance::Dylan"], "conclusion": "may_enter_lab::Dylan", "text": "If a staff member has security clearance and has completed safety training, then that staff member may enter Lab A."}], "query_atom": "may_enter_lab::Dylan", "closure_atoms": ["completed_training::Dylan", "has_clearance::Dylan", "may_enter_lab::Dylan"]}, "source_reports": {"agent_a": {"positive": ["Dylan has security clearance."], "negative": []}, "agent_b": {"positive": ["Dylan has completed safety training."], "negative": []}}, "gold_truth_value": "true", "gold_source_used": "both", "gold_closure_handling": "entailed_or_explicit", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_clearance, may_enter_lab.\nAgent B open/incomplete predicates: assigned_to_trial, completed_training, needs_supervisor_review.\n\nPredicate glossary:\n- assigned_to_trial: is assigned to the clinical trial\n- completed_training: has completed safety training\n- has_clearance: has security clearance\n- may_enter_lab: may enter Lab A\n- needs_supervisor_review: needs supervisor review\n\nCoordinator policy rules:\n- If a staff member has security clearance and has completed safety training, then that staff member may enter Lab A.\n\nAgent A report:\nPositive facts:\n- Dylan has security clearance.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Dylan has completed safety training.\nNegative facts:\n- none\n\nTarget atom: may_enter_lab::Dylan.\nTarget statement: Dylan may enter Lab A.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "hospital_access_0027_entailed_closed_conclusion__owa__multi_agent", "source_id": "hospital_access_0027_entailed_closed_conclusion__owa", "base_id": "hospital_access_0027_entailed_closed_conclusion", "split": "train", "domain": "hospital_access", "family": "entailed_closed_conclusion", "subset": "control_entailed", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["assigned_to_trial", "completed_training", "has_clearance", "may_enter_lab", "needs_supervisor_review"], "facts_positive": ["Dylan has completed safety training.", "Dylan has security clearance."], "facts_negative": [], "rules_natural": ["If a staff member has security clearance and has completed safety training, then that staff member may enter Lab A."], "predicate_glossary": {"completed_training": "has completed safety training", "has_clearance": "has security clearance", "assigned_to_trial": "is assigned to the clinical trial", "may_enter_lab": "may enter Lab A", "needs_supervisor_review": "needs supervisor review"}, "vocabulary_predicates": ["assigned_to_trial", "completed_training", "has_clearance", "may_enter_lab", "needs_supervisor_review"], "query_atom": "may_enter_lab::Dylan", "target_statement": "Dylan may enter Lab A.", "gold_reason_type": "entailed_by_fact_or_rule", "symbolic": {"positive_atoms": ["completed_training::Dylan", "has_clearance::Dylan"], "negative_atoms": [], "rules": [{"antecedents": ["completed_training::Dylan", "has_clearance::Dylan"], "conclusion": "may_enter_lab::Dylan", "text": "If a staff member has security clearance and has completed safety training, then that staff member may enter Lab A."}], "query_atom": "may_enter_lab::Dylan", "closure_atoms": ["completed_training::Dylan", "has_clearance::Dylan", "may_enter_lab::Dylan"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Dylan has completed safety training.", "Dylan has security clearance."], "negative": []}}, "gold_truth_value": "true", "gold_source_used": "agent_b", "gold_closure_handling": "entailed_or_explicit", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: assigned_to_trial, completed_training, has_clearance, may_enter_lab, needs_supervisor_review.\n\nPredicate glossary:\n- assigned_to_trial: is assigned to the clinical trial\n- completed_training: has completed safety training\n- has_clearance: has security clearance\n- may_enter_lab: may enter Lab A\n- needs_supervisor_review: needs supervisor review\n\nCoordinator policy rules:\n- If a staff member has security clearance and has completed safety training, then that staff member may enter Lab A.\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Dylan has completed safety training.\n- Dylan has security clearance.\nNegative facts:\n- none\n\nTarget atom: may_enter_lab::Dylan.\nTarget statement: Dylan may enter Lab A.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "hospital_access_0030_entailed_open_conclusion__cwa__multi_agent", "source_id": "hospital_access_0030_entailed_open_conclusion__cwa", "base_id": "hospital_access_0030_entailed_open_conclusion", "split": "train", "domain": "hospital_access", "family": "entailed_open_conclusion", "subset": "control_entailed", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["assigned_to_trial", "completed_training", "has_clearance", "may_enter_lab", "needs_supervisor_review"], "agent_a_complete_predicates": ["assigned_to_trial", "completed_training", "has_clearance", "may_enter_lab", "needs_supervisor_review"], "agent_b_open_predicates": [], "facts_positive": ["Ava is assigned to the clinical trial."], "facts_negative": [], "rules_natural": ["If a staff member is assigned to the clinical trial, then that staff member needs supervisor review."], "predicate_glossary": {"completed_training": "has completed safety training", "has_clearance": "has security clearance", "assigned_to_trial": "is assigned to the clinical trial", "may_enter_lab": "may enter Lab A", "needs_supervisor_review": "needs supervisor review"}, "vocabulary_predicates": ["assigned_to_trial", "completed_training", "has_clearance", "may_enter_lab", "needs_supervisor_review"], "query_atom": "needs_supervisor_review::Ava", "target_statement": "Ava needs supervisor review.", "gold_reason_type": "entailed_by_fact_or_rule", "symbolic": {"positive_atoms": ["assigned_to_trial::Ava"], "negative_atoms": [], "rules": [{"antecedents": ["assigned_to_trial::Ava"], "conclusion": "needs_supervisor_review::Ava", "text": "If a staff member is assigned to the clinical trial, then that staff member needs supervisor review."}], "query_atom": "needs_supervisor_review::Ava", "closure_atoms": ["assigned_to_trial::Ava", "needs_supervisor_review::Ava"]}, "source_reports": {"agent_a": {"positive": ["Ava is assigned to the clinical trial."], "negative": []}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "true", "gold_source_used": "agent_a", "gold_closure_handling": "entailed_or_explicit", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: assigned_to_trial, completed_training, has_clearance, may_enter_lab, needs_supervisor_review.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- assigned_to_trial: is assigned to the clinical trial\n- completed_training: has completed safety training\n- has_clearance: has security clearance\n- may_enter_lab: may enter Lab A\n- needs_supervisor_review: needs supervisor review\n\nCoordinator policy rules:\n- If a staff member is assigned to the clinical trial, then that staff member needs supervisor review.\n\nAgent A report:\nPositive facts:\n- Ava is assigned to the clinical trial.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: needs_supervisor_review::Ava.\nTarget statement: Ava needs supervisor review.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "hospital_access_0030_entailed_open_conclusion__lcwa__multi_agent", "source_id": "hospital_access_0030_entailed_open_conclusion__lcwa", "base_id": "hospital_access_0030_entailed_open_conclusion", "split": "train", "domain": "hospital_access", "family": "entailed_open_conclusion", "subset": "control_entailed", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_clearance", "may_enter_lab"], "agent_a_complete_predicates": ["has_clearance", "may_enter_lab"], "agent_b_open_predicates": ["assigned_to_trial", "completed_training", "needs_supervisor_review"], "facts_positive": ["Ava is assigned to the clinical trial."], "facts_negative": [], "rules_natural": ["If a staff member is assigned to the clinical trial, then that staff member needs supervisor review."], "predicate_glossary": {"completed_training": "has completed safety training", "has_clearance": "has security clearance", "assigned_to_trial": "is assigned to the clinical trial", "may_enter_lab": "may enter Lab A", "needs_supervisor_review": "needs supervisor review"}, "vocabulary_predicates": ["assigned_to_trial", "completed_training", "has_clearance", "may_enter_lab", "needs_supervisor_review"], "query_atom": "needs_supervisor_review::Ava", "target_statement": "Ava needs supervisor review.", "gold_reason_type": "entailed_by_fact_or_rule", "symbolic": {"positive_atoms": ["assigned_to_trial::Ava"], "negative_atoms": [], "rules": [{"antecedents": ["assigned_to_trial::Ava"], "conclusion": "needs_supervisor_review::Ava", "text": "If a staff member is assigned to the clinical trial, then that staff member needs supervisor review."}], "query_atom": "needs_supervisor_review::Ava", "closure_atoms": ["assigned_to_trial::Ava", "needs_supervisor_review::Ava"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Ava is assigned to the clinical trial."], "negative": []}}, "gold_truth_value": "true", "gold_source_used": "agent_b", "gold_closure_handling": "entailed_or_explicit", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_clearance, may_enter_lab.\nAgent B open/incomplete predicates: assigned_to_trial, completed_training, needs_supervisor_review.\n\nPredicate glossary:\n- assigned_to_trial: is assigned to the clinical trial\n- completed_training: has completed safety training\n- has_clearance: has security clearance\n- may_enter_lab: may enter Lab A\n- needs_supervisor_review: needs supervisor review\n\nCoordinator policy rules:\n- If a staff member is assigned to the clinical trial, then that staff member needs supervisor review.\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Ava is assigned to the clinical trial.\nNegative facts:\n- none\n\nTarget atom: needs_supervisor_review::Ava.\nTarget statement: Ava needs supervisor review.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "hospital_access_0030_entailed_open_conclusion__owa__multi_agent", "source_id": "hospital_access_0030_entailed_open_conclusion__owa", "base_id": "hospital_access_0030_entailed_open_conclusion", "split": "train", "domain": "hospital_access", "family": "entailed_open_conclusion", "subset": "control_entailed", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["assigned_to_trial", "completed_training", "has_clearance", "may_enter_lab", "needs_supervisor_review"], "facts_positive": ["Ava is assigned to the clinical trial."], "facts_negative": [], "rules_natural": ["If a staff member is assigned to the clinical trial, then that staff member needs supervisor review."], "predicate_glossary": {"completed_training": "has completed safety training", "has_clearance": "has security clearance", "assigned_to_trial": "is assigned to the clinical trial", "may_enter_lab": "may enter Lab A", "needs_supervisor_review": "needs supervisor review"}, "vocabulary_predicates": ["assigned_to_trial", "completed_training", "has_clearance", "may_enter_lab", "needs_supervisor_review"], "query_atom": "needs_supervisor_review::Ava", "target_statement": "Ava needs supervisor review.", "gold_reason_type": "entailed_by_fact_or_rule", "symbolic": {"positive_atoms": ["assigned_to_trial::Ava"], "negative_atoms": [], "rules": [{"antecedents": ["assigned_to_trial::Ava"], "conclusion": "needs_supervisor_review::Ava", "text": "If a staff member is assigned to the clinical trial, then that staff member needs supervisor review."}], "query_atom": "needs_supervisor_review::Ava", "closure_atoms": ["assigned_to_trial::Ava", "needs_supervisor_review::Ava"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Ava is assigned to the clinical trial."], "negative": []}}, "gold_truth_value": "true", "gold_source_used": "agent_b", "gold_closure_handling": "entailed_or_explicit", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: assigned_to_trial, completed_training, has_clearance, may_enter_lab, needs_supervisor_review.\n\nPredicate glossary:\n- assigned_to_trial: is assigned to the clinical trial\n- completed_training: has completed safety training\n- has_clearance: has security clearance\n- may_enter_lab: may enter Lab A\n- needs_supervisor_review: needs supervisor review\n\nCoordinator policy rules:\n- If a staff member is assigned to the clinical trial, then that staff member needs supervisor review.\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Ava is assigned to the clinical trial.\nNegative facts:\n- none\n\nTarget atom: needs_supervisor_review::Ava.\nTarget statement: Ava needs supervisor review.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "hospital_access_0032_explicit_negative_closed__cwa__multi_agent", "source_id": "hospital_access_0032_explicit_negative_closed__cwa", "base_id": "hospital_access_0032_explicit_negative_closed", "split": "train", "domain": "hospital_access", "family": "explicit_negative_closed", "subset": "control_explicit_negative", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["assigned_to_trial", "completed_training", "has_clearance", "may_enter_lab", "needs_supervisor_review"], "agent_a_complete_predicates": ["assigned_to_trial", "completed_training", "has_clearance", "may_enter_lab", "needs_supervisor_review"], "agent_b_open_predicates": [], "facts_positive": ["Elena has completed safety training."], "facts_negative": ["Cora does not has security clearance."], "rules_natural": [], "predicate_glossary": {"completed_training": "has completed safety training", "has_clearance": "has security clearance", "assigned_to_trial": "is assigned to the clinical trial", "may_enter_lab": "may enter Lab A", "needs_supervisor_review": "needs supervisor review"}, "vocabulary_predicates": ["assigned_to_trial", "completed_training", "has_clearance", "may_enter_lab", "needs_supervisor_review"], "query_atom": "has_clearance::Cora", "target_statement": "Cora has security clearance.", "gold_reason_type": "explicit_negative_fact", "symbolic": {"positive_atoms": ["completed_training::Elena"], "negative_atoms": ["has_clearance::Cora"], "rules": [], "query_atom": "has_clearance::Cora", "closure_atoms": ["completed_training::Elena"]}, "source_reports": {"agent_a": {"positive": ["Elena has completed safety training."], "negative": ["Cora does not has security clearance."]}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "explicit_negative", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: assigned_to_trial, completed_training, has_clearance, may_enter_lab, needs_supervisor_review.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- assigned_to_trial: is assigned to the clinical trial\n- completed_training: has completed safety training\n- has_clearance: has security clearance\n- may_enter_lab: may enter Lab A\n- needs_supervisor_review: needs supervisor review\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- Elena has completed safety training.\nNegative facts:\n- Cora does not has security clearance.\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: has_clearance::Cora.\nTarget statement: Cora has security clearance.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "hospital_access_0032_explicit_negative_closed__lcwa__multi_agent", "source_id": "hospital_access_0032_explicit_negative_closed__lcwa", "base_id": "hospital_access_0032_explicit_negative_closed", "split": "train", "domain": "hospital_access", "family": "explicit_negative_closed", "subset": "control_explicit_negative", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_clearance"], "agent_a_complete_predicates": ["has_clearance"], "agent_b_open_predicates": ["assigned_to_trial", "completed_training", "may_enter_lab", "needs_supervisor_review"], "facts_positive": ["Elena has completed safety training."], "facts_negative": ["Cora does not has security clearance."], "rules_natural": [], "predicate_glossary": {"completed_training": "has completed safety training", "has_clearance": "has security clearance", "assigned_to_trial": "is assigned to the clinical trial", "may_enter_lab": "may enter Lab A", "needs_supervisor_review": "needs supervisor review"}, "vocabulary_predicates": ["assigned_to_trial", "completed_training", "has_clearance", "may_enter_lab", "needs_supervisor_review"], "query_atom": "has_clearance::Cora", "target_statement": "Cora has security clearance.", "gold_reason_type": "explicit_negative_fact", "symbolic": {"positive_atoms": ["completed_training::Elena"], "negative_atoms": ["has_clearance::Cora"], "rules": [], "query_atom": "has_clearance::Cora", "closure_atoms": ["completed_training::Elena"]}, "source_reports": {"agent_a": {"positive": [], "negative": ["Cora does not has security clearance."]}, "agent_b": {"positive": ["Elena has completed safety training."], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "explicit_negative", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_clearance.\nAgent B open/incomplete predicates: assigned_to_trial, completed_training, may_enter_lab, needs_supervisor_review.\n\nPredicate glossary:\n- assigned_to_trial: is assigned to the clinical trial\n- completed_training: has completed safety training\n- has_clearance: has security clearance\n- may_enter_lab: may enter Lab A\n- needs_supervisor_review: needs supervisor review\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- Cora does not has security clearance.\n\nAgent B report:\nPositive facts:\n- Elena has completed safety training.\nNegative facts:\n- none\n\nTarget atom: has_clearance::Cora.\nTarget statement: Cora has security clearance.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "hospital_access_0032_explicit_negative_closed__owa__multi_agent", "source_id": "hospital_access_0032_explicit_negative_closed__owa", "base_id": "hospital_access_0032_explicit_negative_closed", "split": "train", "domain": "hospital_access", "family": "explicit_negative_closed", "subset": "control_explicit_negative", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["assigned_to_trial", "completed_training", "has_clearance", "may_enter_lab", "needs_supervisor_review"], "facts_positive": ["Elena has completed safety training."], "facts_negative": ["Cora does not has security clearance."], "rules_natural": [], "predicate_glossary": {"completed_training": "has completed safety training", "has_clearance": "has security clearance", "assigned_to_trial": "is assigned to the clinical trial", "may_enter_lab": "may enter Lab A", "needs_supervisor_review": "needs supervisor review"}, "vocabulary_predicates": ["assigned_to_trial", "completed_training", "has_clearance", "may_enter_lab", "needs_supervisor_review"], "query_atom": "has_clearance::Cora", "target_statement": "Cora has security clearance.", "gold_reason_type": "explicit_negative_fact", "symbolic": {"positive_atoms": ["completed_training::Elena"], "negative_atoms": ["has_clearance::Cora"], "rules": [], "query_atom": "has_clearance::Cora", "closure_atoms": ["completed_training::Elena"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Elena has completed safety training."], "negative": ["Cora does not has security clearance."]}}, "gold_truth_value": "false", "gold_source_used": "agent_b", "gold_closure_handling": "explicit_negative", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: assigned_to_trial, completed_training, has_clearance, may_enter_lab, needs_supervisor_review.\n\nPredicate glossary:\n- assigned_to_trial: is assigned to the clinical trial\n- completed_training: has completed safety training\n- has_clearance: has security clearance\n- may_enter_lab: may enter Lab A\n- needs_supervisor_review: needs supervisor review\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Elena has completed safety training.\nNegative facts:\n- Cora does not has security clearance.\n\nTarget atom: has_clearance::Cora.\nTarget statement: Cora has security clearance.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "hospital_access_0033_explicit_negative_closed__cwa__multi_agent", "source_id": "hospital_access_0033_explicit_negative_closed__cwa", "base_id": "hospital_access_0033_explicit_negative_closed", "split": "train", "domain": "hospital_access", "family": "explicit_negative_closed", "subset": "control_explicit_negative", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["assigned_to_trial", "completed_training", "has_clearance", "may_enter_lab", "needs_supervisor_review"], "agent_a_complete_predicates": ["assigned_to_trial", "completed_training", "has_clearance", "may_enter_lab", "needs_supervisor_review"], "agent_b_open_predicates": [], "facts_positive": ["Farid has completed safety training."], "facts_negative": ["Dylan does not has security clearance."], "rules_natural": [], "predicate_glossary": {"completed_training": "has completed safety training", "has_clearance": "has security clearance", "assigned_to_trial": "is assigned to the clinical trial", "may_enter_lab": "may enter Lab A", "needs_supervisor_review": "needs supervisor review"}, "vocabulary_predicates": ["assigned_to_trial", "completed_training", "has_clearance", "may_enter_lab", "needs_supervisor_review"], "query_atom": "has_clearance::Dylan", "target_statement": "Dylan has security clearance.", "gold_reason_type": "explicit_negative_fact", "symbolic": {"positive_atoms": ["completed_training::Farid"], "negative_atoms": ["has_clearance::Dylan"], "rules": [], "query_atom": "has_clearance::Dylan", "closure_atoms": ["completed_training::Farid"]}, "source_reports": {"agent_a": {"positive": ["Farid has completed safety training."], "negative": ["Dylan does not has security clearance."]}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "explicit_negative", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: assigned_to_trial, completed_training, has_clearance, may_enter_lab, needs_supervisor_review.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- assigned_to_trial: is assigned to the clinical trial\n- completed_training: has completed safety training\n- has_clearance: has security clearance\n- may_enter_lab: may enter Lab A\n- needs_supervisor_review: needs supervisor review\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- Farid has completed safety training.\nNegative facts:\n- Dylan does not has security clearance.\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: has_clearance::Dylan.\nTarget statement: Dylan has security clearance.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "hospital_access_0033_explicit_negative_closed__lcwa__multi_agent", "source_id": "hospital_access_0033_explicit_negative_closed__lcwa", "base_id": "hospital_access_0033_explicit_negative_closed", "split": "train", "domain": "hospital_access", "family": "explicit_negative_closed", "subset": "control_explicit_negative", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_clearance"], "agent_a_complete_predicates": ["has_clearance"], "agent_b_open_predicates": ["assigned_to_trial", "completed_training", "may_enter_lab", "needs_supervisor_review"], "facts_positive": ["Farid has completed safety training."], "facts_negative": ["Dylan does not has security clearance."], "rules_natural": [], "predicate_glossary": {"completed_training": "has completed safety training", "has_clearance": "has security clearance", "assigned_to_trial": "is assigned to the clinical trial", "may_enter_lab": "may enter Lab A", "needs_supervisor_review": "needs supervisor review"}, "vocabulary_predicates": ["assigned_to_trial", "completed_training", "has_clearance", "may_enter_lab", "needs_supervisor_review"], "query_atom": "has_clearance::Dylan", "target_statement": "Dylan has security clearance.", "gold_reason_type": "explicit_negative_fact", "symbolic": {"positive_atoms": ["completed_training::Farid"], "negative_atoms": ["has_clearance::Dylan"], "rules": [], "query_atom": "has_clearance::Dylan", "closure_atoms": ["completed_training::Farid"]}, "source_reports": {"agent_a": {"positive": [], "negative": ["Dylan does not has security clearance."]}, "agent_b": {"positive": ["Farid has completed safety training."], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "explicit_negative", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_clearance.\nAgent B open/incomplete predicates: assigned_to_trial, completed_training, may_enter_lab, needs_supervisor_review.\n\nPredicate glossary:\n- assigned_to_trial: is assigned to the clinical trial\n- completed_training: has completed safety training\n- has_clearance: has security clearance\n- may_enter_lab: may enter Lab A\n- needs_supervisor_review: needs supervisor review\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- Dylan does not has security clearance.\n\nAgent B report:\nPositive facts:\n- Farid has completed safety training.\nNegative facts:\n- none\n\nTarget atom: has_clearance::Dylan.\nTarget statement: Dylan has security clearance.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "hospital_access_0033_explicit_negative_closed__owa__multi_agent", "source_id": "hospital_access_0033_explicit_negative_closed__owa", "base_id": "hospital_access_0033_explicit_negative_closed", "split": "train", "domain": "hospital_access", "family": "explicit_negative_closed", "subset": "control_explicit_negative", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["assigned_to_trial", "completed_training", "has_clearance", "may_enter_lab", "needs_supervisor_review"], "facts_positive": ["Farid has completed safety training."], "facts_negative": ["Dylan does not has security clearance."], "rules_natural": [], "predicate_glossary": {"completed_training": "has completed safety training", "has_clearance": "has security clearance", "assigned_to_trial": "is assigned to the clinical trial", "may_enter_lab": "may enter Lab A", "needs_supervisor_review": "needs supervisor review"}, "vocabulary_predicates": ["assigned_to_trial", "completed_training", "has_clearance", "may_enter_lab", "needs_supervisor_review"], "query_atom": "has_clearance::Dylan", "target_statement": "Dylan has security clearance.", "gold_reason_type": "explicit_negative_fact", "symbolic": {"positive_atoms": ["completed_training::Farid"], "negative_atoms": ["has_clearance::Dylan"], "rules": [], "query_atom": "has_clearance::Dylan", "closure_atoms": ["completed_training::Farid"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Farid has completed safety training."], "negative": ["Dylan does not has security clearance."]}}, "gold_truth_value": "false", "gold_source_used": "agent_b", "gold_closure_handling": "explicit_negative", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: assigned_to_trial, completed_training, has_clearance, may_enter_lab, needs_supervisor_review.\n\nPredicate glossary:\n- assigned_to_trial: is assigned to the clinical trial\n- completed_training: has completed safety training\n- has_clearance: has security clearance\n- may_enter_lab: may enter Lab A\n- needs_supervisor_review: needs supervisor review\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Farid has completed safety training.\nNegative facts:\n- Dylan does not has security clearance.\n\nTarget atom: has_clearance::Dylan.\nTarget statement: Dylan has security clearance.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "hospital_access_0035_explicit_negative_closed__cwa__multi_agent", "source_id": "hospital_access_0035_explicit_negative_closed__cwa", "base_id": "hospital_access_0035_explicit_negative_closed", "split": "train", "domain": "hospital_access", "family": "explicit_negative_closed", "subset": "control_explicit_negative", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["assigned_to_trial", "completed_training", "has_clearance", "may_enter_lab", "needs_supervisor_review"], "agent_a_complete_predicates": ["assigned_to_trial", "completed_training", "has_clearance", "may_enter_lab", "needs_supervisor_review"], "agent_b_open_predicates": [], "facts_positive": ["Ben has completed safety training."], "facts_negative": ["Farid does not has security clearance."], "rules_natural": [], "predicate_glossary": {"completed_training": "has completed safety training", "has_clearance": "has security clearance", "assigned_to_trial": "is assigned to the clinical trial", "may_enter_lab": "may enter Lab A", "needs_supervisor_review": "needs supervisor review"}, "vocabulary_predicates": ["assigned_to_trial", "completed_training", "has_clearance", "may_enter_lab", "needs_supervisor_review"], "query_atom": "has_clearance::Farid", "target_statement": "Farid has security clearance.", "gold_reason_type": "explicit_negative_fact", "symbolic": {"positive_atoms": ["completed_training::Ben"], "negative_atoms": ["has_clearance::Farid"], "rules": [], "query_atom": "has_clearance::Farid", "closure_atoms": ["completed_training::Ben"]}, "source_reports": {"agent_a": {"positive": ["Ben has completed safety training."], "negative": ["Farid does not has security clearance."]}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "explicit_negative", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: assigned_to_trial, completed_training, has_clearance, may_enter_lab, needs_supervisor_review.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- assigned_to_trial: is assigned to the clinical trial\n- completed_training: has completed safety training\n- has_clearance: has security clearance\n- may_enter_lab: may enter Lab A\n- needs_supervisor_review: needs supervisor review\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- Ben has completed safety training.\nNegative facts:\n- Farid does not has security clearance.\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: has_clearance::Farid.\nTarget statement: Farid has security clearance.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "hospital_access_0035_explicit_negative_closed__lcwa__multi_agent", "source_id": "hospital_access_0035_explicit_negative_closed__lcwa", "base_id": "hospital_access_0035_explicit_negative_closed", "split": "train", "domain": "hospital_access", "family": "explicit_negative_closed", "subset": "control_explicit_negative", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_clearance"], "agent_a_complete_predicates": ["has_clearance"], "agent_b_open_predicates": ["assigned_to_trial", "completed_training", "may_enter_lab", "needs_supervisor_review"], "facts_positive": ["Ben has completed safety training."], "facts_negative": ["Farid does not has security clearance."], "rules_natural": [], "predicate_glossary": {"completed_training": "has completed safety training", "has_clearance": "has security clearance", "assigned_to_trial": "is assigned to the clinical trial", "may_enter_lab": "may enter Lab A", "needs_supervisor_review": "needs supervisor review"}, "vocabulary_predicates": ["assigned_to_trial", "completed_training", "has_clearance", "may_enter_lab", "needs_supervisor_review"], "query_atom": "has_clearance::Farid", "target_statement": "Farid has security clearance.", "gold_reason_type": "explicit_negative_fact", "symbolic": {"positive_atoms": ["completed_training::Ben"], "negative_atoms": ["has_clearance::Farid"], "rules": [], "query_atom": "has_clearance::Farid", "closure_atoms": ["completed_training::Ben"]}, "source_reports": {"agent_a": {"positive": [], "negative": ["Farid does not has security clearance."]}, "agent_b": {"positive": ["Ben has completed safety training."], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "explicit_negative", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_clearance.\nAgent B open/incomplete predicates: assigned_to_trial, completed_training, may_enter_lab, needs_supervisor_review.\n\nPredicate glossary:\n- assigned_to_trial: is assigned to the clinical trial\n- completed_training: has completed safety training\n- has_clearance: has security clearance\n- may_enter_lab: may enter Lab A\n- needs_supervisor_review: needs supervisor review\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- Farid does not has security clearance.\n\nAgent B report:\nPositive facts:\n- Ben has completed safety training.\nNegative facts:\n- none\n\nTarget atom: has_clearance::Farid.\nTarget statement: Farid has security clearance.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "hospital_access_0035_explicit_negative_closed__owa__multi_agent", "source_id": "hospital_access_0035_explicit_negative_closed__owa", "base_id": "hospital_access_0035_explicit_negative_closed", "split": "train", "domain": "hospital_access", "family": "explicit_negative_closed", "subset": "control_explicit_negative", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["assigned_to_trial", "completed_training", "has_clearance", "may_enter_lab", "needs_supervisor_review"], "facts_positive": ["Ben has completed safety training."], "facts_negative": ["Farid does not has security clearance."], "rules_natural": [], "predicate_glossary": {"completed_training": "has completed safety training", "has_clearance": "has security clearance", "assigned_to_trial": "is assigned to the clinical trial", "may_enter_lab": "may enter Lab A", "needs_supervisor_review": "needs supervisor review"}, "vocabulary_predicates": ["assigned_to_trial", "completed_training", "has_clearance", "may_enter_lab", "needs_supervisor_review"], "query_atom": "has_clearance::Farid", "target_statement": "Farid has security clearance.", "gold_reason_type": "explicit_negative_fact", "symbolic": {"positive_atoms": ["completed_training::Ben"], "negative_atoms": ["has_clearance::Farid"], "rules": [], "query_atom": "has_clearance::Farid", "closure_atoms": ["completed_training::Ben"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Ben has completed safety training."], "negative": ["Farid does not has security clearance."]}}, "gold_truth_value": "false", "gold_source_used": "agent_b", "gold_closure_handling": "explicit_negative", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: assigned_to_trial, completed_training, has_clearance, may_enter_lab, needs_supervisor_review.\n\nPredicate glossary:\n- assigned_to_trial: is assigned to the clinical trial\n- completed_training: has completed safety training\n- has_clearance: has security clearance\n- may_enter_lab: may enter Lab A\n- needs_supervisor_review: needs supervisor review\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Ben has completed safety training.\nNegative facts:\n- Farid does not has security clearance.\n\nTarget atom: has_clearance::Farid.\nTarget statement: Farid has security clearance.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "library_services_0203_closed_missing_direct__cwa__multi_agent", "source_id": "library_services_0203_closed_missing_direct__cwa", "base_id": "library_services_0203_closed_missing_direct", "split": "train", "domain": "library_services", "family": "closed_missing_direct", "subset": "core_contrastive", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["has_active_membership", "has_borrowing_clearance", "has_special_request", "may_borrow_archive_item", "needs_librarian_review"], "agent_a_complete_predicates": ["has_active_membership", "has_borrowing_clearance", "has_special_request", "may_borrow_archive_item", "needs_librarian_review"], "agent_b_open_predicates": [], "facts_positive": ["Reader F has an active membership."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"has_active_membership": "has an active membership", "has_borrowing_clearance": "has borrowing clearance", "has_special_request": "has a special request", "may_borrow_archive_item": "may borrow an archive item", "needs_librarian_review": "needs librarian review"}, "vocabulary_predicates": ["has_active_membership", "has_borrowing_clearance", "has_special_request", "may_borrow_archive_item", "needs_librarian_review"], "query_atom": "has_borrowing_clearance::Reader F", "target_statement": "Reader F has borrowing clearance.", "gold_reason_type": "global_closed_world_underivable", "symbolic": {"positive_atoms": ["has_active_membership::Reader F"], "negative_atoms": [], "rules": [], "query_atom": "has_borrowing_clearance::Reader F", "closure_atoms": ["has_active_membership::Reader F"]}, "source_reports": {"agent_a": {"positive": ["Reader F has an active membership."], "negative": []}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "closed_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_active_membership, has_borrowing_clearance, has_special_request, may_borrow_archive_item, needs_librarian_review.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- has_active_membership: has an active membership\n- has_borrowing_clearance: has borrowing clearance\n- has_special_request: has a special request\n- may_borrow_archive_item: may borrow an archive item\n- needs_librarian_review: needs librarian review\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- Reader F has an active membership.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: has_borrowing_clearance::Reader F.\nTarget statement: Reader F has borrowing clearance.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "library_services_0203_closed_missing_direct__lcwa__multi_agent", "source_id": "library_services_0203_closed_missing_direct__lcwa", "base_id": "library_services_0203_closed_missing_direct", "split": "train", "domain": "library_services", "family": "closed_missing_direct", "subset": "core_contrastive", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_borrowing_clearance"], "agent_a_complete_predicates": ["has_borrowing_clearance"], "agent_b_open_predicates": ["has_active_membership", "has_special_request", "may_borrow_archive_item", "needs_librarian_review"], "facts_positive": ["Reader F has an active membership."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"has_active_membership": "has an active membership", "has_borrowing_clearance": "has borrowing clearance", "has_special_request": "has a special request", "may_borrow_archive_item": "may borrow an archive item", "needs_librarian_review": "needs librarian review"}, "vocabulary_predicates": ["has_active_membership", "has_borrowing_clearance", "has_special_request", "may_borrow_archive_item", "needs_librarian_review"], "query_atom": "has_borrowing_clearance::Reader F", "target_statement": "Reader F has borrowing clearance.", "gold_reason_type": "local_closed_world_underivable", "symbolic": {"positive_atoms": ["has_active_membership::Reader F"], "negative_atoms": [], "rules": [], "query_atom": "has_borrowing_clearance::Reader F", "closure_atoms": ["has_active_membership::Reader F"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Reader F has an active membership."], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "closed_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_borrowing_clearance.\nAgent B open/incomplete predicates: has_active_membership, has_special_request, may_borrow_archive_item, needs_librarian_review.\n\nPredicate glossary:\n- has_active_membership: has an active membership\n- has_borrowing_clearance: has borrowing clearance\n- has_special_request: has a special request\n- may_borrow_archive_item: may borrow an archive item\n- needs_librarian_review: needs librarian review\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Reader F has an active membership.\nNegative facts:\n- none\n\nTarget atom: has_borrowing_clearance::Reader F.\nTarget statement: Reader F has borrowing clearance.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "library_services_0203_closed_missing_direct__owa__multi_agent", "source_id": "library_services_0203_closed_missing_direct__owa", "base_id": "library_services_0203_closed_missing_direct", "split": "train", "domain": "library_services", "family": "closed_missing_direct", "subset": "core_contrastive", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["has_active_membership", "has_borrowing_clearance", "has_special_request", "may_borrow_archive_item", "needs_librarian_review"], "facts_positive": ["Reader F has an active membership."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"has_active_membership": "has an active membership", "has_borrowing_clearance": "has borrowing clearance", "has_special_request": "has a special request", "may_borrow_archive_item": "may borrow an archive item", "needs_librarian_review": "needs librarian review"}, "vocabulary_predicates": ["has_active_membership", "has_borrowing_clearance", "has_special_request", "may_borrow_archive_item", "needs_librarian_review"], "query_atom": "has_borrowing_clearance::Reader F", "target_statement": "Reader F has borrowing clearance.", "gold_reason_type": "open_world_underivable", "symbolic": {"positive_atoms": ["has_active_membership::Reader F"], "negative_atoms": [], "rules": [], "query_atom": "has_borrowing_clearance::Reader F", "closure_atoms": ["has_active_membership::Reader F"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Reader F has an active membership."], "negative": []}}, "gold_truth_value": "unknown", "gold_source_used": "agent_b", "gold_closure_handling": "open_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: has_active_membership, has_borrowing_clearance, has_special_request, may_borrow_archive_item, needs_librarian_review.\n\nPredicate glossary:\n- has_active_membership: has an active membership\n- has_borrowing_clearance: has borrowing clearance\n- has_special_request: has a special request\n- may_borrow_archive_item: may borrow an archive item\n- needs_librarian_review: needs librarian review\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Reader F has an active membership.\nNegative facts:\n- none\n\nTarget atom: has_borrowing_clearance::Reader F.\nTarget statement: Reader F has borrowing clearance.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "library_services_0207_open_missing_direct__cwa__multi_agent", "source_id": "library_services_0207_open_missing_direct__cwa", "base_id": "library_services_0207_open_missing_direct", "split": "train", "domain": "library_services", "family": "open_missing_direct", "subset": "core_contrastive", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["has_active_membership", "has_borrowing_clearance", "has_special_request", "may_borrow_archive_item", "needs_librarian_review"], "agent_a_complete_predicates": ["has_active_membership", "has_borrowing_clearance", "has_special_request", "may_borrow_archive_item", "needs_librarian_review"], "agent_b_open_predicates": [], "facts_positive": ["Reader D has an active membership."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"has_active_membership": "has an active membership", "has_borrowing_clearance": "has borrowing clearance", "has_special_request": "has a special request", "may_borrow_archive_item": "may borrow an archive item", "needs_librarian_review": "needs librarian review"}, "vocabulary_predicates": ["has_active_membership", "has_borrowing_clearance", "has_special_request", "may_borrow_archive_item", "needs_librarian_review"], "query_atom": "has_special_request::Reader D", "target_statement": "Reader D has a special request.", "gold_reason_type": "global_closed_world_underivable", "symbolic": {"positive_atoms": ["has_active_membership::Reader D"], "negative_atoms": [], "rules": [], "query_atom": "has_special_request::Reader D", "closure_atoms": ["has_active_membership::Reader D"]}, "source_reports": {"agent_a": {"positive": ["Reader D has an active membership."], "negative": []}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "closed_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_active_membership, has_borrowing_clearance, has_special_request, may_borrow_archive_item, needs_librarian_review.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- has_active_membership: has an active membership\n- has_borrowing_clearance: has borrowing clearance\n- has_special_request: has a special request\n- may_borrow_archive_item: may borrow an archive item\n- needs_librarian_review: needs librarian review\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- Reader D has an active membership.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: has_special_request::Reader D.\nTarget statement: Reader D has a special request.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "library_services_0207_open_missing_direct__lcwa__multi_agent", "source_id": "library_services_0207_open_missing_direct__lcwa", "base_id": "library_services_0207_open_missing_direct", "split": "train", "domain": "library_services", "family": "open_missing_direct", "subset": "core_contrastive", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_borrowing_clearance", "may_borrow_archive_item"], "agent_a_complete_predicates": ["has_borrowing_clearance", "may_borrow_archive_item"], "agent_b_open_predicates": ["has_active_membership", "has_special_request", "needs_librarian_review"], "facts_positive": ["Reader D has an active membership."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"has_active_membership": "has an active membership", "has_borrowing_clearance": "has borrowing clearance", "has_special_request": "has a special request", "may_borrow_archive_item": "may borrow an archive item", "needs_librarian_review": "needs librarian review"}, "vocabulary_predicates": ["has_active_membership", "has_borrowing_clearance", "has_special_request", "may_borrow_archive_item", "needs_librarian_review"], "query_atom": "has_special_request::Reader D", "target_statement": "Reader D has a special request.", "gold_reason_type": "local_open_world_underivable", "symbolic": {"positive_atoms": ["has_active_membership::Reader D"], "negative_atoms": [], "rules": [], "query_atom": "has_special_request::Reader D", "closure_atoms": ["has_active_membership::Reader D"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Reader D has an active membership."], "negative": []}}, "gold_truth_value": "unknown", "gold_source_used": "agent_b", "gold_closure_handling": "open_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_borrowing_clearance, may_borrow_archive_item.\nAgent B open/incomplete predicates: has_active_membership, has_special_request, needs_librarian_review.\n\nPredicate glossary:\n- has_active_membership: has an active membership\n- has_borrowing_clearance: has borrowing clearance\n- has_special_request: has a special request\n- may_borrow_archive_item: may borrow an archive item\n- needs_librarian_review: needs librarian review\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Reader D has an active membership.\nNegative facts:\n- none\n\nTarget atom: has_special_request::Reader D.\nTarget statement: Reader D has a special request.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "library_services_0207_open_missing_direct__owa__multi_agent", "source_id": "library_services_0207_open_missing_direct__owa", "base_id": "library_services_0207_open_missing_direct", "split": "train", "domain": "library_services", "family": "open_missing_direct", "subset": "core_contrastive", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["has_active_membership", "has_borrowing_clearance", "has_special_request", "may_borrow_archive_item", "needs_librarian_review"], "facts_positive": ["Reader D has an active membership."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"has_active_membership": "has an active membership", "has_borrowing_clearance": "has borrowing clearance", "has_special_request": "has a special request", "may_borrow_archive_item": "may borrow an archive item", "needs_librarian_review": "needs librarian review"}, "vocabulary_predicates": ["has_active_membership", "has_borrowing_clearance", "has_special_request", "may_borrow_archive_item", "needs_librarian_review"], "query_atom": "has_special_request::Reader D", "target_statement": "Reader D has a special request.", "gold_reason_type": "open_world_underivable", "symbolic": {"positive_atoms": ["has_active_membership::Reader D"], "negative_atoms": [], "rules": [], "query_atom": "has_special_request::Reader D", "closure_atoms": ["has_active_membership::Reader D"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Reader D has an active membership."], "negative": []}}, "gold_truth_value": "unknown", "gold_source_used": "agent_b", "gold_closure_handling": "open_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: has_active_membership, has_borrowing_clearance, has_special_request, may_borrow_archive_item, needs_librarian_review.\n\nPredicate glossary:\n- has_active_membership: has an active membership\n- has_borrowing_clearance: has borrowing clearance\n- has_special_request: has a special request\n- may_borrow_archive_item: may borrow an archive item\n- needs_librarian_review: needs librarian review\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Reader D has an active membership.\nNegative facts:\n- none\n\nTarget atom: has_special_request::Reader D.\nTarget statement: Reader D has a special request.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "library_services_0210_closed_missing_with_open_distractor__cwa__multi_agent", "source_id": "library_services_0210_closed_missing_with_open_distractor__cwa", "base_id": "library_services_0210_closed_missing_with_open_distractor", "split": "train", "domain": "library_services", "family": "closed_missing_with_open_distractor", "subset": "core_contrastive", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["has_active_membership", "has_borrowing_clearance", "has_special_request", "may_borrow_archive_item", "needs_librarian_review"], "agent_a_complete_predicates": ["has_active_membership", "has_borrowing_clearance", "has_special_request", "may_borrow_archive_item", "needs_librarian_review"], "agent_b_open_predicates": [], "facts_positive": ["Reader A has an active membership.", "Reader B has a special request."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"has_active_membership": "has an active membership", "has_borrowing_clearance": "has borrowing clearance", "has_special_request": "has a special request", "may_borrow_archive_item": "may borrow an archive item", "needs_librarian_review": "needs librarian review"}, "vocabulary_predicates": ["has_active_membership", "has_borrowing_clearance", "has_special_request", "may_borrow_archive_item", "needs_librarian_review"], "query_atom": "has_borrowing_clearance::Reader A", "target_statement": "Reader A has borrowing clearance.", "gold_reason_type": "global_closed_world_underivable", "symbolic": {"positive_atoms": ["has_active_membership::Reader A", "has_special_request::Reader B"], "negative_atoms": [], "rules": [], "query_atom": "has_borrowing_clearance::Reader A", "closure_atoms": ["has_active_membership::Reader A", "has_special_request::Reader B"]}, "source_reports": {"agent_a": {"positive": ["Reader A has an active membership.", "Reader B has a special request."], "negative": []}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "closed_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_active_membership, has_borrowing_clearance, has_special_request, may_borrow_archive_item, needs_librarian_review.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- has_active_membership: has an active membership\n- has_borrowing_clearance: has borrowing clearance\n- has_special_request: has a special request\n- may_borrow_archive_item: may borrow an archive item\n- needs_librarian_review: needs librarian review\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- Reader A has an active membership.\n- Reader B has a special request.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: has_borrowing_clearance::Reader A.\nTarget statement: Reader A has borrowing clearance.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "library_services_0210_closed_missing_with_open_distractor__lcwa__multi_agent", "source_id": "library_services_0210_closed_missing_with_open_distractor__lcwa", "base_id": "library_services_0210_closed_missing_with_open_distractor", "split": "train", "domain": "library_services", "family": "closed_missing_with_open_distractor", "subset": "core_contrastive", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_borrowing_clearance"], "agent_a_complete_predicates": ["has_borrowing_clearance"], "agent_b_open_predicates": ["has_active_membership", "has_special_request", "may_borrow_archive_item", "needs_librarian_review"], "facts_positive": ["Reader A has an active membership.", "Reader B has a special request."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"has_active_membership": "has an active membership", "has_borrowing_clearance": "has borrowing clearance", "has_special_request": "has a special request", "may_borrow_archive_item": "may borrow an archive item", "needs_librarian_review": "needs librarian review"}, "vocabulary_predicates": ["has_active_membership", "has_borrowing_clearance", "has_special_request", "may_borrow_archive_item", "needs_librarian_review"], "query_atom": "has_borrowing_clearance::Reader A", "target_statement": "Reader A has borrowing clearance.", "gold_reason_type": "local_closed_world_underivable", "symbolic": {"positive_atoms": ["has_active_membership::Reader A", "has_special_request::Reader B"], "negative_atoms": [], "rules": [], "query_atom": "has_borrowing_clearance::Reader A", "closure_atoms": ["has_active_membership::Reader A", "has_special_request::Reader B"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Reader A has an active membership.", "Reader B has a special request."], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "closed_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_borrowing_clearance.\nAgent B open/incomplete predicates: has_active_membership, has_special_request, may_borrow_archive_item, needs_librarian_review.\n\nPredicate glossary:\n- has_active_membership: has an active membership\n- has_borrowing_clearance: has borrowing clearance\n- has_special_request: has a special request\n- may_borrow_archive_item: may borrow an archive item\n- needs_librarian_review: needs librarian review\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Reader A has an active membership.\n- Reader B has a special request.\nNegative facts:\n- none\n\nTarget atom: has_borrowing_clearance::Reader A.\nTarget statement: Reader A has borrowing clearance.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "library_services_0210_closed_missing_with_open_distractor__owa__multi_agent", "source_id": "library_services_0210_closed_missing_with_open_distractor__owa", "base_id": "library_services_0210_closed_missing_with_open_distractor", "split": "train", "domain": "library_services", "family": "closed_missing_with_open_distractor", "subset": "core_contrastive", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["has_active_membership", "has_borrowing_clearance", "has_special_request", "may_borrow_archive_item", "needs_librarian_review"], "facts_positive": ["Reader A has an active membership.", "Reader B has a special request."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"has_active_membership": "has an active membership", "has_borrowing_clearance": "has borrowing clearance", "has_special_request": "has a special request", "may_borrow_archive_item": "may borrow an archive item", "needs_librarian_review": "needs librarian review"}, "vocabulary_predicates": ["has_active_membership", "has_borrowing_clearance", "has_special_request", "may_borrow_archive_item", "needs_librarian_review"], "query_atom": "has_borrowing_clearance::Reader A", "target_statement": "Reader A has borrowing clearance.", "gold_reason_type": "open_world_underivable", "symbolic": {"positive_atoms": ["has_active_membership::Reader A", "has_special_request::Reader B"], "negative_atoms": [], "rules": [], "query_atom": "has_borrowing_clearance::Reader A", "closure_atoms": ["has_active_membership::Reader A", "has_special_request::Reader B"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Reader A has an active membership.", "Reader B has a special request."], "negative": []}}, "gold_truth_value": "unknown", "gold_source_used": "agent_b", "gold_closure_handling": "open_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: has_active_membership, has_borrowing_clearance, has_special_request, may_borrow_archive_item, needs_librarian_review.\n\nPredicate glossary:\n- has_active_membership: has an active membership\n- has_borrowing_clearance: has borrowing clearance\n- has_special_request: has a special request\n- may_borrow_archive_item: may borrow an archive item\n- needs_librarian_review: needs librarian review\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Reader A has an active membership.\n- Reader B has a special request.\nNegative facts:\n- none\n\nTarget atom: has_borrowing_clearance::Reader A.\nTarget statement: Reader A has borrowing clearance.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "library_services_0211_closed_missing_with_open_distractor__cwa__multi_agent", "source_id": "library_services_0211_closed_missing_with_open_distractor__cwa", "base_id": "library_services_0211_closed_missing_with_open_distractor", "split": "test", "domain": "library_services", "family": "closed_missing_with_open_distractor", "subset": "core_contrastive", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["has_active_membership", "has_borrowing_clearance", "has_special_request", "may_borrow_archive_item", "needs_librarian_review"], "agent_a_complete_predicates": ["has_active_membership", "has_borrowing_clearance", "has_special_request", "may_borrow_archive_item", "needs_librarian_review"], "agent_b_open_predicates": [], "facts_positive": ["Reader B has an active membership.", "Reader C has a special request."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"has_active_membership": "has an active membership", "has_borrowing_clearance": "has borrowing clearance", "has_special_request": "has a special request", "may_borrow_archive_item": "may borrow an archive item", "needs_librarian_review": "needs librarian review"}, "vocabulary_predicates": ["has_active_membership", "has_borrowing_clearance", "has_special_request", "may_borrow_archive_item", "needs_librarian_review"], "query_atom": "has_borrowing_clearance::Reader B", "target_statement": "Reader B has borrowing clearance.", "gold_reason_type": "global_closed_world_underivable", "symbolic": {"positive_atoms": ["has_active_membership::Reader B", "has_special_request::Reader C"], "negative_atoms": [], "rules": [], "query_atom": "has_borrowing_clearance::Reader B", "closure_atoms": ["has_active_membership::Reader B", "has_special_request::Reader C"]}, "source_reports": {"agent_a": {"positive": ["Reader B has an active membership.", "Reader C has a special request."], "negative": []}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "closed_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_active_membership, has_borrowing_clearance, has_special_request, may_borrow_archive_item, needs_librarian_review.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- has_active_membership: has an active membership\n- has_borrowing_clearance: has borrowing clearance\n- has_special_request: has a special request\n- may_borrow_archive_item: may borrow an archive item\n- needs_librarian_review: needs librarian review\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- Reader B has an active membership.\n- Reader C has a special request.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: has_borrowing_clearance::Reader B.\nTarget statement: Reader B has borrowing clearance.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "library_services_0211_closed_missing_with_open_distractor__lcwa__multi_agent", "source_id": "library_services_0211_closed_missing_with_open_distractor__lcwa", "base_id": "library_services_0211_closed_missing_with_open_distractor", "split": "test", "domain": "library_services", "family": "closed_missing_with_open_distractor", "subset": "core_contrastive", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_borrowing_clearance"], "agent_a_complete_predicates": ["has_borrowing_clearance"], "agent_b_open_predicates": ["has_active_membership", "has_special_request", "may_borrow_archive_item", "needs_librarian_review"], "facts_positive": ["Reader B has an active membership.", "Reader C has a special request."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"has_active_membership": "has an active membership", "has_borrowing_clearance": "has borrowing clearance", "has_special_request": "has a special request", "may_borrow_archive_item": "may borrow an archive item", "needs_librarian_review": "needs librarian review"}, "vocabulary_predicates": ["has_active_membership", "has_borrowing_clearance", "has_special_request", "may_borrow_archive_item", "needs_librarian_review"], "query_atom": "has_borrowing_clearance::Reader B", "target_statement": "Reader B has borrowing clearance.", "gold_reason_type": "local_closed_world_underivable", "symbolic": {"positive_atoms": ["has_active_membership::Reader B", "has_special_request::Reader C"], "negative_atoms": [], "rules": [], "query_atom": "has_borrowing_clearance::Reader B", "closure_atoms": ["has_active_membership::Reader B", "has_special_request::Reader C"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Reader B has an active membership.", "Reader C has a special request."], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "closed_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_borrowing_clearance.\nAgent B open/incomplete predicates: has_active_membership, has_special_request, may_borrow_archive_item, needs_librarian_review.\n\nPredicate glossary:\n- has_active_membership: has an active membership\n- has_borrowing_clearance: has borrowing clearance\n- has_special_request: has a special request\n- may_borrow_archive_item: may borrow an archive item\n- needs_librarian_review: needs librarian review\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Reader B has an active membership.\n- Reader C has a special request.\nNegative facts:\n- none\n\nTarget atom: has_borrowing_clearance::Reader B.\nTarget statement: Reader B has borrowing clearance.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "library_services_0211_closed_missing_with_open_distractor__owa__multi_agent", "source_id": "library_services_0211_closed_missing_with_open_distractor__owa", "base_id": "library_services_0211_closed_missing_with_open_distractor", "split": "test", "domain": "library_services", "family": "closed_missing_with_open_distractor", "subset": "core_contrastive", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["has_active_membership", "has_borrowing_clearance", "has_special_request", "may_borrow_archive_item", "needs_librarian_review"], "facts_positive": ["Reader B has an active membership.", "Reader C has a special request."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"has_active_membership": "has an active membership", "has_borrowing_clearance": "has borrowing clearance", "has_special_request": "has a special request", "may_borrow_archive_item": "may borrow an archive item", "needs_librarian_review": "needs librarian review"}, "vocabulary_predicates": ["has_active_membership", "has_borrowing_clearance", "has_special_request", "may_borrow_archive_item", "needs_librarian_review"], "query_atom": "has_borrowing_clearance::Reader B", "target_statement": "Reader B has borrowing clearance.", "gold_reason_type": "open_world_underivable", "symbolic": {"positive_atoms": ["has_active_membership::Reader B", "has_special_request::Reader C"], "negative_atoms": [], "rules": [], "query_atom": "has_borrowing_clearance::Reader B", "closure_atoms": ["has_active_membership::Reader B", "has_special_request::Reader C"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Reader B has an active membership.", "Reader C has a special request."], "negative": []}}, "gold_truth_value": "unknown", "gold_source_used": "agent_b", "gold_closure_handling": "open_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: has_active_membership, has_borrowing_clearance, has_special_request, may_borrow_archive_item, needs_librarian_review.\n\nPredicate glossary:\n- has_active_membership: has an active membership\n- has_borrowing_clearance: has borrowing clearance\n- has_special_request: has a special request\n- may_borrow_archive_item: may borrow an archive item\n- needs_librarian_review: needs librarian review\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Reader B has an active membership.\n- Reader C has a special request.\nNegative facts:\n- none\n\nTarget atom: has_borrowing_clearance::Reader B.\nTarget statement: Reader B has borrowing clearance.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "library_services_0220_open_derived_missing_antecedent__cwa__multi_agent", "source_id": "library_services_0220_open_derived_missing_antecedent__cwa", "base_id": "library_services_0220_open_derived_missing_antecedent", "split": "test", "domain": "library_services", "family": "open_derived_missing_antecedent", "subset": "core_contrastive", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["has_active_membership", "has_borrowing_clearance", "has_special_request", "may_borrow_archive_item", "needs_librarian_review"], "agent_a_complete_predicates": ["has_active_membership", "has_borrowing_clearance", "has_special_request", "may_borrow_archive_item", "needs_librarian_review"], "agent_b_open_predicates": [], "facts_positive": ["Reader E has an active membership."], "facts_negative": [], "rules_natural": ["If a reader has a special request, then that reader needs librarian review."], "predicate_glossary": {"has_active_membership": "has an active membership", "has_borrowing_clearance": "has borrowing clearance", "has_special_request": "has a special request", "may_borrow_archive_item": "may borrow an archive item", "needs_librarian_review": "needs librarian review"}, "vocabulary_predicates": ["has_active_membership", "has_borrowing_clearance", "has_special_request", "may_borrow_archive_item", "needs_librarian_review"], "query_atom": "needs_librarian_review::Reader E", "target_statement": "Reader E needs librarian review.", "gold_reason_type": "global_closed_world_underivable", "symbolic": {"positive_atoms": ["has_active_membership::Reader E"], "negative_atoms": [], "rules": [{"antecedents": ["has_special_request::Reader E"], "conclusion": "needs_librarian_review::Reader E", "text": "If a reader has a special request, then that reader needs librarian review."}], "query_atom": "needs_librarian_review::Reader E", "closure_atoms": ["has_active_membership::Reader E"]}, "source_reports": {"agent_a": {"positive": ["Reader E has an active membership."], "negative": []}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "closed_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_active_membership, has_borrowing_clearance, has_special_request, may_borrow_archive_item, needs_librarian_review.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- has_active_membership: has an active membership\n- has_borrowing_clearance: has borrowing clearance\n- has_special_request: has a special request\n- may_borrow_archive_item: may borrow an archive item\n- needs_librarian_review: needs librarian review\n\nCoordinator policy rules:\n- If a reader has a special request, then that reader needs librarian review.\n\nAgent A report:\nPositive facts:\n- Reader E has an active membership.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: needs_librarian_review::Reader E.\nTarget statement: Reader E needs librarian review.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "library_services_0220_open_derived_missing_antecedent__lcwa__multi_agent", "source_id": "library_services_0220_open_derived_missing_antecedent__lcwa", "base_id": "library_services_0220_open_derived_missing_antecedent", "split": "test", "domain": "library_services", "family": "open_derived_missing_antecedent", "subset": "core_contrastive", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_borrowing_clearance", "may_borrow_archive_item"], "agent_a_complete_predicates": ["has_borrowing_clearance", "may_borrow_archive_item"], "agent_b_open_predicates": ["has_active_membership", "has_special_request", "needs_librarian_review"], "facts_positive": ["Reader E has an active membership."], "facts_negative": [], "rules_natural": ["If a reader has a special request, then that reader needs librarian review."], "predicate_glossary": {"has_active_membership": "has an active membership", "has_borrowing_clearance": "has borrowing clearance", "has_special_request": "has a special request", "may_borrow_archive_item": "may borrow an archive item", "needs_librarian_review": "needs librarian review"}, "vocabulary_predicates": ["has_active_membership", "has_borrowing_clearance", "has_special_request", "may_borrow_archive_item", "needs_librarian_review"], "query_atom": "needs_librarian_review::Reader E", "target_statement": "Reader E needs librarian review.", "gold_reason_type": "local_open_world_underivable", "symbolic": {"positive_atoms": ["has_active_membership::Reader E"], "negative_atoms": [], "rules": [{"antecedents": ["has_special_request::Reader E"], "conclusion": "needs_librarian_review::Reader E", "text": "If a reader has a special request, then that reader needs librarian review."}], "query_atom": "needs_librarian_review::Reader E", "closure_atoms": ["has_active_membership::Reader E"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Reader E has an active membership."], "negative": []}}, "gold_truth_value": "unknown", "gold_source_used": "agent_b", "gold_closure_handling": "open_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_borrowing_clearance, may_borrow_archive_item.\nAgent B open/incomplete predicates: has_active_membership, has_special_request, needs_librarian_review.\n\nPredicate glossary:\n- has_active_membership: has an active membership\n- has_borrowing_clearance: has borrowing clearance\n- has_special_request: has a special request\n- may_borrow_archive_item: may borrow an archive item\n- needs_librarian_review: needs librarian review\n\nCoordinator policy rules:\n- If a reader has a special request, then that reader needs librarian review.\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Reader E has an active membership.\nNegative facts:\n- none\n\nTarget atom: needs_librarian_review::Reader E.\nTarget statement: Reader E needs librarian review.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "library_services_0220_open_derived_missing_antecedent__owa__multi_agent", "source_id": "library_services_0220_open_derived_missing_antecedent__owa", "base_id": "library_services_0220_open_derived_missing_antecedent", "split": "test", "domain": "library_services", "family": "open_derived_missing_antecedent", "subset": "core_contrastive", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["has_active_membership", "has_borrowing_clearance", "has_special_request", "may_borrow_archive_item", "needs_librarian_review"], "facts_positive": ["Reader E has an active membership."], "facts_negative": [], "rules_natural": ["If a reader has a special request, then that reader needs librarian review."], "predicate_glossary": {"has_active_membership": "has an active membership", "has_borrowing_clearance": "has borrowing clearance", "has_special_request": "has a special request", "may_borrow_archive_item": "may borrow an archive item", "needs_librarian_review": "needs librarian review"}, "vocabulary_predicates": ["has_active_membership", "has_borrowing_clearance", "has_special_request", "may_borrow_archive_item", "needs_librarian_review"], "query_atom": "needs_librarian_review::Reader E", "target_statement": "Reader E needs librarian review.", "gold_reason_type": "open_world_underivable", "symbolic": {"positive_atoms": ["has_active_membership::Reader E"], "negative_atoms": [], "rules": [{"antecedents": ["has_special_request::Reader E"], "conclusion": "needs_librarian_review::Reader E", "text": "If a reader has a special request, then that reader needs librarian review."}], "query_atom": "needs_librarian_review::Reader E", "closure_atoms": ["has_active_membership::Reader E"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Reader E has an active membership."], "negative": []}}, "gold_truth_value": "unknown", "gold_source_used": "agent_b", "gold_closure_handling": "open_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: has_active_membership, has_borrowing_clearance, has_special_request, may_borrow_archive_item, needs_librarian_review.\n\nPredicate glossary:\n- has_active_membership: has an active membership\n- has_borrowing_clearance: has borrowing clearance\n- has_special_request: has a special request\n- may_borrow_archive_item: may borrow an archive item\n- needs_librarian_review: needs librarian review\n\nCoordinator policy rules:\n- If a reader has a special request, then that reader needs librarian review.\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Reader E has an active membership.\nNegative facts:\n- none\n\nTarget atom: needs_librarian_review::Reader E.\nTarget statement: Reader E needs librarian review.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "library_services_0226_entailed_closed_conclusion__cwa__multi_agent", "source_id": "library_services_0226_entailed_closed_conclusion__cwa", "base_id": "library_services_0226_entailed_closed_conclusion", "split": "train", "domain": "library_services", "family": "entailed_closed_conclusion", "subset": "control_entailed", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["has_active_membership", "has_borrowing_clearance", "has_special_request", "may_borrow_archive_item", "needs_librarian_review"], "agent_a_complete_predicates": ["has_active_membership", "has_borrowing_clearance", "has_special_request", "may_borrow_archive_item", "needs_librarian_review"], "agent_b_open_predicates": [], "facts_positive": ["Reader E has an active membership.", "Reader E has borrowing clearance."], "facts_negative": [], "rules_natural": ["If a reader has an active membership and has borrowing clearance, then that reader may borrow an archive item."], "predicate_glossary": {"has_active_membership": "has an active membership", "has_borrowing_clearance": "has borrowing clearance", "has_special_request": "has a special request", "may_borrow_archive_item": "may borrow an archive item", "needs_librarian_review": "needs librarian review"}, "vocabulary_predicates": ["has_active_membership", "has_borrowing_clearance", "has_special_request", "may_borrow_archive_item", "needs_librarian_review"], "query_atom": "may_borrow_archive_item::Reader E", "target_statement": "Reader E may borrow an archive item.", "gold_reason_type": "entailed_by_fact_or_rule", "symbolic": {"positive_atoms": ["has_active_membership::Reader E", "has_borrowing_clearance::Reader E"], "negative_atoms": [], "rules": [{"antecedents": ["has_active_membership::Reader E", "has_borrowing_clearance::Reader E"], "conclusion": "may_borrow_archive_item::Reader E", "text": "If a reader has an active membership and has borrowing clearance, then that reader may borrow an archive item."}], "query_atom": "may_borrow_archive_item::Reader E", "closure_atoms": ["has_active_membership::Reader E", "has_borrowing_clearance::Reader E", "may_borrow_archive_item::Reader E"]}, "source_reports": {"agent_a": {"positive": ["Reader E has an active membership.", "Reader E has borrowing clearance."], "negative": []}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "true", "gold_source_used": "agent_a", "gold_closure_handling": "entailed_or_explicit", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_active_membership, has_borrowing_clearance, has_special_request, may_borrow_archive_item, needs_librarian_review.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- has_active_membership: has an active membership\n- has_borrowing_clearance: has borrowing clearance\n- has_special_request: has a special request\n- may_borrow_archive_item: may borrow an archive item\n- needs_librarian_review: needs librarian review\n\nCoordinator policy rules:\n- If a reader has an active membership and has borrowing clearance, then that reader may borrow an archive item.\n\nAgent A report:\nPositive facts:\n- Reader E has an active membership.\n- Reader E has borrowing clearance.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: may_borrow_archive_item::Reader E.\nTarget statement: Reader E may borrow an archive item.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "library_services_0226_entailed_closed_conclusion__lcwa__multi_agent", "source_id": "library_services_0226_entailed_closed_conclusion__lcwa", "base_id": "library_services_0226_entailed_closed_conclusion", "split": "train", "domain": "library_services", "family": "entailed_closed_conclusion", "subset": "control_entailed", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_borrowing_clearance", "may_borrow_archive_item"], "agent_a_complete_predicates": ["has_borrowing_clearance", "may_borrow_archive_item"], "agent_b_open_predicates": ["has_active_membership", "has_special_request", "needs_librarian_review"], "facts_positive": ["Reader E has an active membership.", "Reader E has borrowing clearance."], "facts_negative": [], "rules_natural": ["If a reader has an active membership and has borrowing clearance, then that reader may borrow an archive item."], "predicate_glossary": {"has_active_membership": "has an active membership", "has_borrowing_clearance": "has borrowing clearance", "has_special_request": "has a special request", "may_borrow_archive_item": "may borrow an archive item", "needs_librarian_review": "needs librarian review"}, "vocabulary_predicates": ["has_active_membership", "has_borrowing_clearance", "has_special_request", "may_borrow_archive_item", "needs_librarian_review"], "query_atom": "may_borrow_archive_item::Reader E", "target_statement": "Reader E may borrow an archive item.", "gold_reason_type": "entailed_by_fact_or_rule", "symbolic": {"positive_atoms": ["has_active_membership::Reader E", "has_borrowing_clearance::Reader E"], "negative_atoms": [], "rules": [{"antecedents": ["has_active_membership::Reader E", "has_borrowing_clearance::Reader E"], "conclusion": "may_borrow_archive_item::Reader E", "text": "If a reader has an active membership and has borrowing clearance, then that reader may borrow an archive item."}], "query_atom": "may_borrow_archive_item::Reader E", "closure_atoms": ["has_active_membership::Reader E", "has_borrowing_clearance::Reader E", "may_borrow_archive_item::Reader E"]}, "source_reports": {"agent_a": {"positive": ["Reader E has borrowing clearance."], "negative": []}, "agent_b": {"positive": ["Reader E has an active membership."], "negative": []}}, "gold_truth_value": "true", "gold_source_used": "both", "gold_closure_handling": "entailed_or_explicit", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_borrowing_clearance, may_borrow_archive_item.\nAgent B open/incomplete predicates: has_active_membership, has_special_request, needs_librarian_review.\n\nPredicate glossary:\n- has_active_membership: has an active membership\n- has_borrowing_clearance: has borrowing clearance\n- has_special_request: has a special request\n- may_borrow_archive_item: may borrow an archive item\n- needs_librarian_review: needs librarian review\n\nCoordinator policy rules:\n- If a reader has an active membership and has borrowing clearance, then that reader may borrow an archive item.\n\nAgent A report:\nPositive facts:\n- Reader E has borrowing clearance.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Reader E has an active membership.\nNegative facts:\n- none\n\nTarget atom: may_borrow_archive_item::Reader E.\nTarget statement: Reader E may borrow an archive item.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "library_services_0226_entailed_closed_conclusion__owa__multi_agent", "source_id": "library_services_0226_entailed_closed_conclusion__owa", "base_id": "library_services_0226_entailed_closed_conclusion", "split": "train", "domain": "library_services", "family": "entailed_closed_conclusion", "subset": "control_entailed", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["has_active_membership", "has_borrowing_clearance", "has_special_request", "may_borrow_archive_item", "needs_librarian_review"], "facts_positive": ["Reader E has an active membership.", "Reader E has borrowing clearance."], "facts_negative": [], "rules_natural": ["If a reader has an active membership and has borrowing clearance, then that reader may borrow an archive item."], "predicate_glossary": {"has_active_membership": "has an active membership", "has_borrowing_clearance": "has borrowing clearance", "has_special_request": "has a special request", "may_borrow_archive_item": "may borrow an archive item", "needs_librarian_review": "needs librarian review"}, "vocabulary_predicates": ["has_active_membership", "has_borrowing_clearance", "has_special_request", "may_borrow_archive_item", "needs_librarian_review"], "query_atom": "may_borrow_archive_item::Reader E", "target_statement": "Reader E may borrow an archive item.", "gold_reason_type": "entailed_by_fact_or_rule", "symbolic": {"positive_atoms": ["has_active_membership::Reader E", "has_borrowing_clearance::Reader E"], "negative_atoms": [], "rules": [{"antecedents": ["has_active_membership::Reader E", "has_borrowing_clearance::Reader E"], "conclusion": "may_borrow_archive_item::Reader E", "text": "If a reader has an active membership and has borrowing clearance, then that reader may borrow an archive item."}], "query_atom": "may_borrow_archive_item::Reader E", "closure_atoms": ["has_active_membership::Reader E", "has_borrowing_clearance::Reader E", "may_borrow_archive_item::Reader E"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Reader E has an active membership.", "Reader E has borrowing clearance."], "negative": []}}, "gold_truth_value": "true", "gold_source_used": "agent_b", "gold_closure_handling": "entailed_or_explicit", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: has_active_membership, has_borrowing_clearance, has_special_request, may_borrow_archive_item, needs_librarian_review.\n\nPredicate glossary:\n- has_active_membership: has an active membership\n- has_borrowing_clearance: has borrowing clearance\n- has_special_request: has a special request\n- may_borrow_archive_item: may borrow an archive item\n- needs_librarian_review: needs librarian review\n\nCoordinator policy rules:\n- If a reader has an active membership and has borrowing clearance, then that reader may borrow an archive item.\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Reader E has an active membership.\n- Reader E has borrowing clearance.\nNegative facts:\n- none\n\nTarget atom: may_borrow_archive_item::Reader E.\nTarget statement: Reader E may borrow an archive item.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "library_services_0227_entailed_closed_conclusion__cwa__multi_agent", "source_id": "library_services_0227_entailed_closed_conclusion__cwa", "base_id": "library_services_0227_entailed_closed_conclusion", "split": "train", "domain": "library_services", "family": "entailed_closed_conclusion", "subset": "control_entailed", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["has_active_membership", "has_borrowing_clearance", "has_special_request", "may_borrow_archive_item", "needs_librarian_review"], "agent_a_complete_predicates": ["has_active_membership", "has_borrowing_clearance", "has_special_request", "may_borrow_archive_item", "needs_librarian_review"], "agent_b_open_predicates": [], "facts_positive": ["Reader F has an active membership.", "Reader F has borrowing clearance."], "facts_negative": [], "rules_natural": ["If a reader has an active membership and has borrowing clearance, then that reader may borrow an archive item."], "predicate_glossary": {"has_active_membership": "has an active membership", "has_borrowing_clearance": "has borrowing clearance", "has_special_request": "has a special request", "may_borrow_archive_item": "may borrow an archive item", "needs_librarian_review": "needs librarian review"}, "vocabulary_predicates": ["has_active_membership", "has_borrowing_clearance", "has_special_request", "may_borrow_archive_item", "needs_librarian_review"], "query_atom": "may_borrow_archive_item::Reader F", "target_statement": "Reader F may borrow an archive item.", "gold_reason_type": "entailed_by_fact_or_rule", "symbolic": {"positive_atoms": ["has_active_membership::Reader F", "has_borrowing_clearance::Reader F"], "negative_atoms": [], "rules": [{"antecedents": ["has_active_membership::Reader F", "has_borrowing_clearance::Reader F"], "conclusion": "may_borrow_archive_item::Reader F", "text": "If a reader has an active membership and has borrowing clearance, then that reader may borrow an archive item."}], "query_atom": "may_borrow_archive_item::Reader F", "closure_atoms": ["has_active_membership::Reader F", "has_borrowing_clearance::Reader F", "may_borrow_archive_item::Reader F"]}, "source_reports": {"agent_a": {"positive": ["Reader F has an active membership.", "Reader F has borrowing clearance."], "negative": []}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "true", "gold_source_used": "agent_a", "gold_closure_handling": "entailed_or_explicit", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_active_membership, has_borrowing_clearance, has_special_request, may_borrow_archive_item, needs_librarian_review.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- has_active_membership: has an active membership\n- has_borrowing_clearance: has borrowing clearance\n- has_special_request: has a special request\n- may_borrow_archive_item: may borrow an archive item\n- needs_librarian_review: needs librarian review\n\nCoordinator policy rules:\n- If a reader has an active membership and has borrowing clearance, then that reader may borrow an archive item.\n\nAgent A report:\nPositive facts:\n- Reader F has an active membership.\n- Reader F has borrowing clearance.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: may_borrow_archive_item::Reader F.\nTarget statement: Reader F may borrow an archive item.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "library_services_0227_entailed_closed_conclusion__lcwa__multi_agent", "source_id": "library_services_0227_entailed_closed_conclusion__lcwa", "base_id": "library_services_0227_entailed_closed_conclusion", "split": "train", "domain": "library_services", "family": "entailed_closed_conclusion", "subset": "control_entailed", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_borrowing_clearance", "may_borrow_archive_item"], "agent_a_complete_predicates": ["has_borrowing_clearance", "may_borrow_archive_item"], "agent_b_open_predicates": ["has_active_membership", "has_special_request", "needs_librarian_review"], "facts_positive": ["Reader F has an active membership.", "Reader F has borrowing clearance."], "facts_negative": [], "rules_natural": ["If a reader has an active membership and has borrowing clearance, then that reader may borrow an archive item."], "predicate_glossary": {"has_active_membership": "has an active membership", "has_borrowing_clearance": "has borrowing clearance", "has_special_request": "has a special request", "may_borrow_archive_item": "may borrow an archive item", "needs_librarian_review": "needs librarian review"}, "vocabulary_predicates": ["has_active_membership", "has_borrowing_clearance", "has_special_request", "may_borrow_archive_item", "needs_librarian_review"], "query_atom": "may_borrow_archive_item::Reader F", "target_statement": "Reader F may borrow an archive item.", "gold_reason_type": "entailed_by_fact_or_rule", "symbolic": {"positive_atoms": ["has_active_membership::Reader F", "has_borrowing_clearance::Reader F"], "negative_atoms": [], "rules": [{"antecedents": ["has_active_membership::Reader F", "has_borrowing_clearance::Reader F"], "conclusion": "may_borrow_archive_item::Reader F", "text": "If a reader has an active membership and has borrowing clearance, then that reader may borrow an archive item."}], "query_atom": "may_borrow_archive_item::Reader F", "closure_atoms": ["has_active_membership::Reader F", "has_borrowing_clearance::Reader F", "may_borrow_archive_item::Reader F"]}, "source_reports": {"agent_a": {"positive": ["Reader F has borrowing clearance."], "negative": []}, "agent_b": {"positive": ["Reader F has an active membership."], "negative": []}}, "gold_truth_value": "true", "gold_source_used": "both", "gold_closure_handling": "entailed_or_explicit", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_borrowing_clearance, may_borrow_archive_item.\nAgent B open/incomplete predicates: has_active_membership, has_special_request, needs_librarian_review.\n\nPredicate glossary:\n- has_active_membership: has an active membership\n- has_borrowing_clearance: has borrowing clearance\n- has_special_request: has a special request\n- may_borrow_archive_item: may borrow an archive item\n- needs_librarian_review: needs librarian review\n\nCoordinator policy rules:\n- If a reader has an active membership and has borrowing clearance, then that reader may borrow an archive item.\n\nAgent A report:\nPositive facts:\n- Reader F has borrowing clearance.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Reader F has an active membership.\nNegative facts:\n- none\n\nTarget atom: may_borrow_archive_item::Reader F.\nTarget statement: Reader F may borrow an archive item.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "library_services_0227_entailed_closed_conclusion__owa__multi_agent", "source_id": "library_services_0227_entailed_closed_conclusion__owa", "base_id": "library_services_0227_entailed_closed_conclusion", "split": "train", "domain": "library_services", "family": "entailed_closed_conclusion", "subset": "control_entailed", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["has_active_membership", "has_borrowing_clearance", "has_special_request", "may_borrow_archive_item", "needs_librarian_review"], "facts_positive": ["Reader F has an active membership.", "Reader F has borrowing clearance."], "facts_negative": [], "rules_natural": ["If a reader has an active membership and has borrowing clearance, then that reader may borrow an archive item."], "predicate_glossary": {"has_active_membership": "has an active membership", "has_borrowing_clearance": "has borrowing clearance", "has_special_request": "has a special request", "may_borrow_archive_item": "may borrow an archive item", "needs_librarian_review": "needs librarian review"}, "vocabulary_predicates": ["has_active_membership", "has_borrowing_clearance", "has_special_request", "may_borrow_archive_item", "needs_librarian_review"], "query_atom": "may_borrow_archive_item::Reader F", "target_statement": "Reader F may borrow an archive item.", "gold_reason_type": "entailed_by_fact_or_rule", "symbolic": {"positive_atoms": ["has_active_membership::Reader F", "has_borrowing_clearance::Reader F"], "negative_atoms": [], "rules": [{"antecedents": ["has_active_membership::Reader F", "has_borrowing_clearance::Reader F"], "conclusion": "may_borrow_archive_item::Reader F", "text": "If a reader has an active membership and has borrowing clearance, then that reader may borrow an archive item."}], "query_atom": "may_borrow_archive_item::Reader F", "closure_atoms": ["has_active_membership::Reader F", "has_borrowing_clearance::Reader F", "may_borrow_archive_item::Reader F"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Reader F has an active membership.", "Reader F has borrowing clearance."], "negative": []}}, "gold_truth_value": "true", "gold_source_used": "agent_b", "gold_closure_handling": "entailed_or_explicit", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: has_active_membership, has_borrowing_clearance, has_special_request, may_borrow_archive_item, needs_librarian_review.\n\nPredicate glossary:\n- has_active_membership: has an active membership\n- has_borrowing_clearance: has borrowing clearance\n- has_special_request: has a special request\n- may_borrow_archive_item: may borrow an archive item\n- needs_librarian_review: needs librarian review\n\nCoordinator policy rules:\n- If a reader has an active membership and has borrowing clearance, then that reader may borrow an archive item.\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Reader F has an active membership.\n- Reader F has borrowing clearance.\nNegative facts:\n- none\n\nTarget atom: may_borrow_archive_item::Reader F.\nTarget statement: Reader F may borrow an archive item.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "library_services_0228_entailed_open_conclusion__cwa__multi_agent", "source_id": "library_services_0228_entailed_open_conclusion__cwa", "base_id": "library_services_0228_entailed_open_conclusion", "split": "dev", "domain": "library_services", "family": "entailed_open_conclusion", "subset": "control_entailed", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["has_active_membership", "has_borrowing_clearance", "has_special_request", "may_borrow_archive_item", "needs_librarian_review"], "agent_a_complete_predicates": ["has_active_membership", "has_borrowing_clearance", "has_special_request", "may_borrow_archive_item", "needs_librarian_review"], "agent_b_open_predicates": [], "facts_positive": ["Reader A has a special request."], "facts_negative": [], "rules_natural": ["If a reader has a special request, then that reader needs librarian review."], "predicate_glossary": {"has_active_membership": "has an active membership", "has_borrowing_clearance": "has borrowing clearance", "has_special_request": "has a special request", "may_borrow_archive_item": "may borrow an archive item", "needs_librarian_review": "needs librarian review"}, "vocabulary_predicates": ["has_active_membership", "has_borrowing_clearance", "has_special_request", "may_borrow_archive_item", "needs_librarian_review"], "query_atom": "needs_librarian_review::Reader A", "target_statement": "Reader A needs librarian review.", "gold_reason_type": "entailed_by_fact_or_rule", "symbolic": {"positive_atoms": ["has_special_request::Reader A"], "negative_atoms": [], "rules": [{"antecedents": ["has_special_request::Reader A"], "conclusion": "needs_librarian_review::Reader A", "text": "If a reader has a special request, then that reader needs librarian review."}], "query_atom": "needs_librarian_review::Reader A", "closure_atoms": ["has_special_request::Reader A", "needs_librarian_review::Reader A"]}, "source_reports": {"agent_a": {"positive": ["Reader A has a special request."], "negative": []}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "true", "gold_source_used": "agent_a", "gold_closure_handling": "entailed_or_explicit", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_active_membership, has_borrowing_clearance, has_special_request, may_borrow_archive_item, needs_librarian_review.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- has_active_membership: has an active membership\n- has_borrowing_clearance: has borrowing clearance\n- has_special_request: has a special request\n- may_borrow_archive_item: may borrow an archive item\n- needs_librarian_review: needs librarian review\n\nCoordinator policy rules:\n- If a reader has a special request, then that reader needs librarian review.\n\nAgent A report:\nPositive facts:\n- Reader A has a special request.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: needs_librarian_review::Reader A.\nTarget statement: Reader A needs librarian review.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "library_services_0228_entailed_open_conclusion__lcwa__multi_agent", "source_id": "library_services_0228_entailed_open_conclusion__lcwa", "base_id": "library_services_0228_entailed_open_conclusion", "split": "dev", "domain": "library_services", "family": "entailed_open_conclusion", "subset": "control_entailed", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_borrowing_clearance", "may_borrow_archive_item"], "agent_a_complete_predicates": ["has_borrowing_clearance", "may_borrow_archive_item"], "agent_b_open_predicates": ["has_active_membership", "has_special_request", "needs_librarian_review"], "facts_positive": ["Reader A has a special request."], "facts_negative": [], "rules_natural": ["If a reader has a special request, then that reader needs librarian review."], "predicate_glossary": {"has_active_membership": "has an active membership", "has_borrowing_clearance": "has borrowing clearance", "has_special_request": "has a special request", "may_borrow_archive_item": "may borrow an archive item", "needs_librarian_review": "needs librarian review"}, "vocabulary_predicates": ["has_active_membership", "has_borrowing_clearance", "has_special_request", "may_borrow_archive_item", "needs_librarian_review"], "query_atom": "needs_librarian_review::Reader A", "target_statement": "Reader A needs librarian review.", "gold_reason_type": "entailed_by_fact_or_rule", "symbolic": {"positive_atoms": ["has_special_request::Reader A"], "negative_atoms": [], "rules": [{"antecedents": ["has_special_request::Reader A"], "conclusion": "needs_librarian_review::Reader A", "text": "If a reader has a special request, then that reader needs librarian review."}], "query_atom": "needs_librarian_review::Reader A", "closure_atoms": ["has_special_request::Reader A", "needs_librarian_review::Reader A"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Reader A has a special request."], "negative": []}}, "gold_truth_value": "true", "gold_source_used": "agent_b", "gold_closure_handling": "entailed_or_explicit", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_borrowing_clearance, may_borrow_archive_item.\nAgent B open/incomplete predicates: has_active_membership, has_special_request, needs_librarian_review.\n\nPredicate glossary:\n- has_active_membership: has an active membership\n- has_borrowing_clearance: has borrowing clearance\n- has_special_request: has a special request\n- may_borrow_archive_item: may borrow an archive item\n- needs_librarian_review: needs librarian review\n\nCoordinator policy rules:\n- If a reader has a special request, then that reader needs librarian review.\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Reader A has a special request.\nNegative facts:\n- none\n\nTarget atom: needs_librarian_review::Reader A.\nTarget statement: Reader A needs librarian review.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "library_services_0228_entailed_open_conclusion__owa__multi_agent", "source_id": "library_services_0228_entailed_open_conclusion__owa", "base_id": "library_services_0228_entailed_open_conclusion", "split": "dev", "domain": "library_services", "family": "entailed_open_conclusion", "subset": "control_entailed", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["has_active_membership", "has_borrowing_clearance", "has_special_request", "may_borrow_archive_item", "needs_librarian_review"], "facts_positive": ["Reader A has a special request."], "facts_negative": [], "rules_natural": ["If a reader has a special request, then that reader needs librarian review."], "predicate_glossary": {"has_active_membership": "has an active membership", "has_borrowing_clearance": "has borrowing clearance", "has_special_request": "has a special request", "may_borrow_archive_item": "may borrow an archive item", "needs_librarian_review": "needs librarian review"}, "vocabulary_predicates": ["has_active_membership", "has_borrowing_clearance", "has_special_request", "may_borrow_archive_item", "needs_librarian_review"], "query_atom": "needs_librarian_review::Reader A", "target_statement": "Reader A needs librarian review.", "gold_reason_type": "entailed_by_fact_or_rule", "symbolic": {"positive_atoms": ["has_special_request::Reader A"], "negative_atoms": [], "rules": [{"antecedents": ["has_special_request::Reader A"], "conclusion": "needs_librarian_review::Reader A", "text": "If a reader has a special request, then that reader needs librarian review."}], "query_atom": "needs_librarian_review::Reader A", "closure_atoms": ["has_special_request::Reader A", "needs_librarian_review::Reader A"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Reader A has a special request."], "negative": []}}, "gold_truth_value": "true", "gold_source_used": "agent_b", "gold_closure_handling": "entailed_or_explicit", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: has_active_membership, has_borrowing_clearance, has_special_request, may_borrow_archive_item, needs_librarian_review.\n\nPredicate glossary:\n- has_active_membership: has an active membership\n- has_borrowing_clearance: has borrowing clearance\n- has_special_request: has a special request\n- may_borrow_archive_item: may borrow an archive item\n- needs_librarian_review: needs librarian review\n\nCoordinator policy rules:\n- If a reader has a special request, then that reader needs librarian review.\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Reader A has a special request.\nNegative facts:\n- none\n\nTarget atom: needs_librarian_review::Reader A.\nTarget statement: Reader A needs librarian review.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "library_services_0235_explicit_negative_closed__cwa__multi_agent", "source_id": "library_services_0235_explicit_negative_closed__cwa", "base_id": "library_services_0235_explicit_negative_closed", "split": "test", "domain": "library_services", "family": "explicit_negative_closed", "subset": "control_explicit_negative", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["has_active_membership", "has_borrowing_clearance", "has_special_request", "may_borrow_archive_item", "needs_librarian_review"], "agent_a_complete_predicates": ["has_active_membership", "has_borrowing_clearance", "has_special_request", "may_borrow_archive_item", "needs_librarian_review"], "agent_b_open_predicates": [], "facts_positive": ["Reader D has an active membership."], "facts_negative": ["Reader B does not has borrowing clearance."], "rules_natural": [], "predicate_glossary": {"has_active_membership": "has an active membership", "has_borrowing_clearance": "has borrowing clearance", "has_special_request": "has a special request", "may_borrow_archive_item": "may borrow an archive item", "needs_librarian_review": "needs librarian review"}, "vocabulary_predicates": ["has_active_membership", "has_borrowing_clearance", "has_special_request", "may_borrow_archive_item", "needs_librarian_review"], "query_atom": "has_borrowing_clearance::Reader B", "target_statement": "Reader B has borrowing clearance.", "gold_reason_type": "explicit_negative_fact", "symbolic": {"positive_atoms": ["has_active_membership::Reader D"], "negative_atoms": ["has_borrowing_clearance::Reader B"], "rules": [], "query_atom": "has_borrowing_clearance::Reader B", "closure_atoms": ["has_active_membership::Reader D"]}, "source_reports": {"agent_a": {"positive": ["Reader D has an active membership."], "negative": ["Reader B does not has borrowing clearance."]}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "explicit_negative", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_active_membership, has_borrowing_clearance, has_special_request, may_borrow_archive_item, needs_librarian_review.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- has_active_membership: has an active membership\n- has_borrowing_clearance: has borrowing clearance\n- has_special_request: has a special request\n- may_borrow_archive_item: may borrow an archive item\n- needs_librarian_review: needs librarian review\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- Reader D has an active membership.\nNegative facts:\n- Reader B does not has borrowing clearance.\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: has_borrowing_clearance::Reader B.\nTarget statement: Reader B has borrowing clearance.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "library_services_0235_explicit_negative_closed__lcwa__multi_agent", "source_id": "library_services_0235_explicit_negative_closed__lcwa", "base_id": "library_services_0235_explicit_negative_closed", "split": "test", "domain": "library_services", "family": "explicit_negative_closed", "subset": "control_explicit_negative", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_borrowing_clearance"], "agent_a_complete_predicates": ["has_borrowing_clearance"], "agent_b_open_predicates": ["has_active_membership", "has_special_request", "may_borrow_archive_item", "needs_librarian_review"], "facts_positive": ["Reader D has an active membership."], "facts_negative": ["Reader B does not has borrowing clearance."], "rules_natural": [], "predicate_glossary": {"has_active_membership": "has an active membership", "has_borrowing_clearance": "has borrowing clearance", "has_special_request": "has a special request", "may_borrow_archive_item": "may borrow an archive item", "needs_librarian_review": "needs librarian review"}, "vocabulary_predicates": ["has_active_membership", "has_borrowing_clearance", "has_special_request", "may_borrow_archive_item", "needs_librarian_review"], "query_atom": "has_borrowing_clearance::Reader B", "target_statement": "Reader B has borrowing clearance.", "gold_reason_type": "explicit_negative_fact", "symbolic": {"positive_atoms": ["has_active_membership::Reader D"], "negative_atoms": ["has_borrowing_clearance::Reader B"], "rules": [], "query_atom": "has_borrowing_clearance::Reader B", "closure_atoms": ["has_active_membership::Reader D"]}, "source_reports": {"agent_a": {"positive": [], "negative": ["Reader B does not has borrowing clearance."]}, "agent_b": {"positive": ["Reader D has an active membership."], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "explicit_negative", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_borrowing_clearance.\nAgent B open/incomplete predicates: has_active_membership, has_special_request, may_borrow_archive_item, needs_librarian_review.\n\nPredicate glossary:\n- has_active_membership: has an active membership\n- has_borrowing_clearance: has borrowing clearance\n- has_special_request: has a special request\n- may_borrow_archive_item: may borrow an archive item\n- needs_librarian_review: needs librarian review\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- Reader B does not has borrowing clearance.\n\nAgent B report:\nPositive facts:\n- Reader D has an active membership.\nNegative facts:\n- none\n\nTarget atom: has_borrowing_clearance::Reader B.\nTarget statement: Reader B has borrowing clearance.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "library_services_0235_explicit_negative_closed__owa__multi_agent", "source_id": "library_services_0235_explicit_negative_closed__owa", "base_id": "library_services_0235_explicit_negative_closed", "split": "test", "domain": "library_services", "family": "explicit_negative_closed", "subset": "control_explicit_negative", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["has_active_membership", "has_borrowing_clearance", "has_special_request", "may_borrow_archive_item", "needs_librarian_review"], "facts_positive": ["Reader D has an active membership."], "facts_negative": ["Reader B does not has borrowing clearance."], "rules_natural": [], "predicate_glossary": {"has_active_membership": "has an active membership", "has_borrowing_clearance": "has borrowing clearance", "has_special_request": "has a special request", "may_borrow_archive_item": "may borrow an archive item", "needs_librarian_review": "needs librarian review"}, "vocabulary_predicates": ["has_active_membership", "has_borrowing_clearance", "has_special_request", "may_borrow_archive_item", "needs_librarian_review"], "query_atom": "has_borrowing_clearance::Reader B", "target_statement": "Reader B has borrowing clearance.", "gold_reason_type": "explicit_negative_fact", "symbolic": {"positive_atoms": ["has_active_membership::Reader D"], "negative_atoms": ["has_borrowing_clearance::Reader B"], "rules": [], "query_atom": "has_borrowing_clearance::Reader B", "closure_atoms": ["has_active_membership::Reader D"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Reader D has an active membership."], "negative": ["Reader B does not has borrowing clearance."]}}, "gold_truth_value": "false", "gold_source_used": "agent_b", "gold_closure_handling": "explicit_negative", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: has_active_membership, has_borrowing_clearance, has_special_request, may_borrow_archive_item, needs_librarian_review.\n\nPredicate glossary:\n- has_active_membership: has an active membership\n- has_borrowing_clearance: has borrowing clearance\n- has_special_request: has a special request\n- may_borrow_archive_item: may borrow an archive item\n- needs_librarian_review: needs librarian review\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Reader D has an active membership.\nNegative facts:\n- Reader B does not has borrowing clearance.\n\nTarget atom: has_borrowing_clearance::Reader B.\nTarget statement: Reader B has borrowing clearance.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "library_services_0237_explicit_negative_open__cwa__multi_agent", "source_id": "library_services_0237_explicit_negative_open__cwa", "base_id": "library_services_0237_explicit_negative_open", "split": "dev", "domain": "library_services", "family": "explicit_negative_open", "subset": "control_explicit_negative", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["has_active_membership", "has_borrowing_clearance", "has_special_request", "may_borrow_archive_item", "needs_librarian_review"], "agent_a_complete_predicates": ["has_active_membership", "has_borrowing_clearance", "has_special_request", "may_borrow_archive_item", "needs_librarian_review"], "agent_b_open_predicates": [], "facts_positive": ["Reader F has an active membership."], "facts_negative": ["Reader D does not has a special request."], "rules_natural": [], "predicate_glossary": {"has_active_membership": "has an active membership", "has_borrowing_clearance": "has borrowing clearance", "has_special_request": "has a special request", "may_borrow_archive_item": "may borrow an archive item", "needs_librarian_review": "needs librarian review"}, "vocabulary_predicates": ["has_active_membership", "has_borrowing_clearance", "has_special_request", "may_borrow_archive_item", "needs_librarian_review"], "query_atom": "has_special_request::Reader D", "target_statement": "Reader D has a special request.", "gold_reason_type": "explicit_negative_fact", "symbolic": {"positive_atoms": ["has_active_membership::Reader F"], "negative_atoms": ["has_special_request::Reader D"], "rules": [], "query_atom": "has_special_request::Reader D", "closure_atoms": ["has_active_membership::Reader F"]}, "source_reports": {"agent_a": {"positive": ["Reader F has an active membership."], "negative": ["Reader D does not has a special request."]}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "explicit_negative", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_active_membership, has_borrowing_clearance, has_special_request, may_borrow_archive_item, needs_librarian_review.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- has_active_membership: has an active membership\n- has_borrowing_clearance: has borrowing clearance\n- has_special_request: has a special request\n- may_borrow_archive_item: may borrow an archive item\n- needs_librarian_review: needs librarian review\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- Reader F has an active membership.\nNegative facts:\n- Reader D does not has a special request.\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: has_special_request::Reader D.\nTarget statement: Reader D has a special request.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "library_services_0237_explicit_negative_open__lcwa__multi_agent", "source_id": "library_services_0237_explicit_negative_open__lcwa", "base_id": "library_services_0237_explicit_negative_open", "split": "dev", "domain": "library_services", "family": "explicit_negative_open", "subset": "control_explicit_negative", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_borrowing_clearance", "may_borrow_archive_item"], "agent_a_complete_predicates": ["has_borrowing_clearance", "may_borrow_archive_item"], "agent_b_open_predicates": ["has_active_membership", "has_special_request", "needs_librarian_review"], "facts_positive": ["Reader F has an active membership."], "facts_negative": ["Reader D does not has a special request."], "rules_natural": [], "predicate_glossary": {"has_active_membership": "has an active membership", "has_borrowing_clearance": "has borrowing clearance", "has_special_request": "has a special request", "may_borrow_archive_item": "may borrow an archive item", "needs_librarian_review": "needs librarian review"}, "vocabulary_predicates": ["has_active_membership", "has_borrowing_clearance", "has_special_request", "may_borrow_archive_item", "needs_librarian_review"], "query_atom": "has_special_request::Reader D", "target_statement": "Reader D has a special request.", "gold_reason_type": "explicit_negative_fact", "symbolic": {"positive_atoms": ["has_active_membership::Reader F"], "negative_atoms": ["has_special_request::Reader D"], "rules": [], "query_atom": "has_special_request::Reader D", "closure_atoms": ["has_active_membership::Reader F"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Reader F has an active membership."], "negative": ["Reader D does not has a special request."]}}, "gold_truth_value": "false", "gold_source_used": "agent_b", "gold_closure_handling": "explicit_negative", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_borrowing_clearance, may_borrow_archive_item.\nAgent B open/incomplete predicates: has_active_membership, has_special_request, needs_librarian_review.\n\nPredicate glossary:\n- has_active_membership: has an active membership\n- has_borrowing_clearance: has borrowing clearance\n- has_special_request: has a special request\n- may_borrow_archive_item: may borrow an archive item\n- needs_librarian_review: needs librarian review\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Reader F has an active membership.\nNegative facts:\n- Reader D does not has a special request.\n\nTarget atom: has_special_request::Reader D.\nTarget statement: Reader D has a special request.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "library_services_0237_explicit_negative_open__owa__multi_agent", "source_id": "library_services_0237_explicit_negative_open__owa", "base_id": "library_services_0237_explicit_negative_open", "split": "dev", "domain": "library_services", "family": "explicit_negative_open", "subset": "control_explicit_negative", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["has_active_membership", "has_borrowing_clearance", "has_special_request", "may_borrow_archive_item", "needs_librarian_review"], "facts_positive": ["Reader F has an active membership."], "facts_negative": ["Reader D does not has a special request."], "rules_natural": [], "predicate_glossary": {"has_active_membership": "has an active membership", "has_borrowing_clearance": "has borrowing clearance", "has_special_request": "has a special request", "may_borrow_archive_item": "may borrow an archive item", "needs_librarian_review": "needs librarian review"}, "vocabulary_predicates": ["has_active_membership", "has_borrowing_clearance", "has_special_request", "may_borrow_archive_item", "needs_librarian_review"], "query_atom": "has_special_request::Reader D", "target_statement": "Reader D has a special request.", "gold_reason_type": "explicit_negative_fact", "symbolic": {"positive_atoms": ["has_active_membership::Reader F"], "negative_atoms": ["has_special_request::Reader D"], "rules": [], "query_atom": "has_special_request::Reader D", "closure_atoms": ["has_active_membership::Reader F"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Reader F has an active membership."], "negative": ["Reader D does not has a special request."]}}, "gold_truth_value": "false", "gold_source_used": "agent_b", "gold_closure_handling": "explicit_negative", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: has_active_membership, has_borrowing_clearance, has_special_request, may_borrow_archive_item, needs_librarian_review.\n\nPredicate glossary:\n- has_active_membership: has an active membership\n- has_borrowing_clearance: has borrowing clearance\n- has_special_request: has a special request\n- may_borrow_archive_item: may borrow an archive item\n- needs_librarian_review: needs librarian review\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Reader F has an active membership.\nNegative facts:\n- Reader D does not has a special request.\n\nTarget atom: has_special_request::Reader D.\nTarget statement: Reader D has a special request.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "manufacturing_quality_0082_closed_missing_direct__cwa__multi_agent", "source_id": "manufacturing_quality_0082_closed_missing_direct__cwa", "base_id": "manufacturing_quality_0082_closed_missing_direct", "split": "train", "domain": "manufacturing_quality", "family": "closed_missing_direct", "subset": "core_contrastive", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "agent_a_complete_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "agent_b_open_predicates": [], "facts_positive": ["Part E passed the visual check."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_visual_check": "passed the visual check", "has_release_approval": "has release approval", "has_supplier_note": "has a supplier note", "may_ship": "may ship to customers", "needs_engineer_review": "needs engineer review"}, "vocabulary_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "query_atom": "has_release_approval::Part E", "target_statement": "Part E has release approval.", "gold_reason_type": "global_closed_world_underivable", "symbolic": {"positive_atoms": ["passed_visual_check::Part E"], "negative_atoms": [], "rules": [], "query_atom": "has_release_approval::Part E", "closure_atoms": ["passed_visual_check::Part E"]}, "source_reports": {"agent_a": {"positive": ["Part E passed the visual check."], "negative": []}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "closed_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_release_approval, has_supplier_note, may_ship, needs_engineer_review, passed_visual_check.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- has_release_approval: has release approval\n- has_supplier_note: has a supplier note\n- may_ship: may ship to customers\n- needs_engineer_review: needs engineer review\n- passed_visual_check: passed the visual check\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- Part E passed the visual check.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: has_release_approval::Part E.\nTarget statement: Part E has release approval.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "manufacturing_quality_0082_closed_missing_direct__lcwa__multi_agent", "source_id": "manufacturing_quality_0082_closed_missing_direct__lcwa", "base_id": "manufacturing_quality_0082_closed_missing_direct", "split": "train", "domain": "manufacturing_quality", "family": "closed_missing_direct", "subset": "core_contrastive", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_release_approval"], "agent_a_complete_predicates": ["has_release_approval"], "agent_b_open_predicates": ["has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "facts_positive": ["Part E passed the visual check."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_visual_check": "passed the visual check", "has_release_approval": "has release approval", "has_supplier_note": "has a supplier note", "may_ship": "may ship to customers", "needs_engineer_review": "needs engineer review"}, "vocabulary_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "query_atom": "has_release_approval::Part E", "target_statement": "Part E has release approval.", "gold_reason_type": "local_closed_world_underivable", "symbolic": {"positive_atoms": ["passed_visual_check::Part E"], "negative_atoms": [], "rules": [], "query_atom": "has_release_approval::Part E", "closure_atoms": ["passed_visual_check::Part E"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Part E passed the visual check."], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "closed_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_release_approval.\nAgent B open/incomplete predicates: has_supplier_note, may_ship, needs_engineer_review, passed_visual_check.\n\nPredicate glossary:\n- has_release_approval: has release approval\n- has_supplier_note: has a supplier note\n- may_ship: may ship to customers\n- needs_engineer_review: needs engineer review\n- passed_visual_check: passed the visual check\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Part E passed the visual check.\nNegative facts:\n- none\n\nTarget atom: has_release_approval::Part E.\nTarget statement: Part E has release approval.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "manufacturing_quality_0082_closed_missing_direct__owa__multi_agent", "source_id": "manufacturing_quality_0082_closed_missing_direct__owa", "base_id": "manufacturing_quality_0082_closed_missing_direct", "split": "train", "domain": "manufacturing_quality", "family": "closed_missing_direct", "subset": "core_contrastive", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "facts_positive": ["Part E passed the visual check."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_visual_check": "passed the visual check", "has_release_approval": "has release approval", "has_supplier_note": "has a supplier note", "may_ship": "may ship to customers", "needs_engineer_review": "needs engineer review"}, "vocabulary_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "query_atom": "has_release_approval::Part E", "target_statement": "Part E has release approval.", "gold_reason_type": "open_world_underivable", "symbolic": {"positive_atoms": ["passed_visual_check::Part E"], "negative_atoms": [], "rules": [], "query_atom": "has_release_approval::Part E", "closure_atoms": ["passed_visual_check::Part E"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Part E passed the visual check."], "negative": []}}, "gold_truth_value": "unknown", "gold_source_used": "agent_b", "gold_closure_handling": "open_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: has_release_approval, has_supplier_note, may_ship, needs_engineer_review, passed_visual_check.\n\nPredicate glossary:\n- has_release_approval: has release approval\n- has_supplier_note: has a supplier note\n- may_ship: may ship to customers\n- needs_engineer_review: needs engineer review\n- passed_visual_check: passed the visual check\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Part E passed the visual check.\nNegative facts:\n- none\n\nTarget atom: has_release_approval::Part E.\nTarget statement: Part E has release approval.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "manufacturing_quality_0085_open_missing_direct__cwa__multi_agent", "source_id": "manufacturing_quality_0085_open_missing_direct__cwa", "base_id": "manufacturing_quality_0085_open_missing_direct", "split": "dev", "domain": "manufacturing_quality", "family": "open_missing_direct", "subset": "core_contrastive", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "agent_a_complete_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "agent_b_open_predicates": [], "facts_positive": ["Part B passed the visual check."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_visual_check": "passed the visual check", "has_release_approval": "has release approval", "has_supplier_note": "has a supplier note", "may_ship": "may ship to customers", "needs_engineer_review": "needs engineer review"}, "vocabulary_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "query_atom": "has_supplier_note::Part B", "target_statement": "Part B has a supplier note.", "gold_reason_type": "global_closed_world_underivable", "symbolic": {"positive_atoms": ["passed_visual_check::Part B"], "negative_atoms": [], "rules": [], "query_atom": "has_supplier_note::Part B", "closure_atoms": ["passed_visual_check::Part B"]}, "source_reports": {"agent_a": {"positive": ["Part B passed the visual check."], "negative": []}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "closed_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_release_approval, has_supplier_note, may_ship, needs_engineer_review, passed_visual_check.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- has_release_approval: has release approval\n- has_supplier_note: has a supplier note\n- may_ship: may ship to customers\n- needs_engineer_review: needs engineer review\n- passed_visual_check: passed the visual check\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- Part B passed the visual check.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: has_supplier_note::Part B.\nTarget statement: Part B has a supplier note.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "manufacturing_quality_0085_open_missing_direct__lcwa__multi_agent", "source_id": "manufacturing_quality_0085_open_missing_direct__lcwa", "base_id": "manufacturing_quality_0085_open_missing_direct", "split": "dev", "domain": "manufacturing_quality", "family": "open_missing_direct", "subset": "core_contrastive", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_release_approval", "may_ship"], "agent_a_complete_predicates": ["has_release_approval", "may_ship"], "agent_b_open_predicates": ["has_supplier_note", "needs_engineer_review", "passed_visual_check"], "facts_positive": ["Part B passed the visual check."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_visual_check": "passed the visual check", "has_release_approval": "has release approval", "has_supplier_note": "has a supplier note", "may_ship": "may ship to customers", "needs_engineer_review": "needs engineer review"}, "vocabulary_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "query_atom": "has_supplier_note::Part B", "target_statement": "Part B has a supplier note.", "gold_reason_type": "local_open_world_underivable", "symbolic": {"positive_atoms": ["passed_visual_check::Part B"], "negative_atoms": [], "rules": [], "query_atom": "has_supplier_note::Part B", "closure_atoms": ["passed_visual_check::Part B"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Part B passed the visual check."], "negative": []}}, "gold_truth_value": "unknown", "gold_source_used": "agent_b", "gold_closure_handling": "open_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_release_approval, may_ship.\nAgent B open/incomplete predicates: has_supplier_note, needs_engineer_review, passed_visual_check.\n\nPredicate glossary:\n- has_release_approval: has release approval\n- has_supplier_note: has a supplier note\n- may_ship: may ship to customers\n- needs_engineer_review: needs engineer review\n- passed_visual_check: passed the visual check\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Part B passed the visual check.\nNegative facts:\n- none\n\nTarget atom: has_supplier_note::Part B.\nTarget statement: Part B has a supplier note.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "manufacturing_quality_0085_open_missing_direct__owa__multi_agent", "source_id": "manufacturing_quality_0085_open_missing_direct__owa", "base_id": "manufacturing_quality_0085_open_missing_direct", "split": "dev", "domain": "manufacturing_quality", "family": "open_missing_direct", "subset": "core_contrastive", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "facts_positive": ["Part B passed the visual check."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_visual_check": "passed the visual check", "has_release_approval": "has release approval", "has_supplier_note": "has a supplier note", "may_ship": "may ship to customers", "needs_engineer_review": "needs engineer review"}, "vocabulary_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "query_atom": "has_supplier_note::Part B", "target_statement": "Part B has a supplier note.", "gold_reason_type": "open_world_underivable", "symbolic": {"positive_atoms": ["passed_visual_check::Part B"], "negative_atoms": [], "rules": [], "query_atom": "has_supplier_note::Part B", "closure_atoms": ["passed_visual_check::Part B"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Part B passed the visual check."], "negative": []}}, "gold_truth_value": "unknown", "gold_source_used": "agent_b", "gold_closure_handling": "open_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: has_release_approval, has_supplier_note, may_ship, needs_engineer_review, passed_visual_check.\n\nPredicate glossary:\n- has_release_approval: has release approval\n- has_supplier_note: has a supplier note\n- may_ship: may ship to customers\n- needs_engineer_review: needs engineer review\n- passed_visual_check: passed the visual check\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Part B passed the visual check.\nNegative facts:\n- none\n\nTarget atom: has_supplier_note::Part B.\nTarget statement: Part B has a supplier note.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "manufacturing_quality_0087_open_missing_direct__cwa__multi_agent", "source_id": "manufacturing_quality_0087_open_missing_direct__cwa", "base_id": "manufacturing_quality_0087_open_missing_direct", "split": "train", "domain": "manufacturing_quality", "family": "open_missing_direct", "subset": "core_contrastive", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "agent_a_complete_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "agent_b_open_predicates": [], "facts_positive": ["Part D passed the visual check."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_visual_check": "passed the visual check", "has_release_approval": "has release approval", "has_supplier_note": "has a supplier note", "may_ship": "may ship to customers", "needs_engineer_review": "needs engineer review"}, "vocabulary_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "query_atom": "has_supplier_note::Part D", "target_statement": "Part D has a supplier note.", "gold_reason_type": "global_closed_world_underivable", "symbolic": {"positive_atoms": ["passed_visual_check::Part D"], "negative_atoms": [], "rules": [], "query_atom": "has_supplier_note::Part D", "closure_atoms": ["passed_visual_check::Part D"]}, "source_reports": {"agent_a": {"positive": ["Part D passed the visual check."], "negative": []}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "closed_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_release_approval, has_supplier_note, may_ship, needs_engineer_review, passed_visual_check.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- has_release_approval: has release approval\n- has_supplier_note: has a supplier note\n- may_ship: may ship to customers\n- needs_engineer_review: needs engineer review\n- passed_visual_check: passed the visual check\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- Part D passed the visual check.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: has_supplier_note::Part D.\nTarget statement: Part D has a supplier note.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "manufacturing_quality_0087_open_missing_direct__lcwa__multi_agent", "source_id": "manufacturing_quality_0087_open_missing_direct__lcwa", "base_id": "manufacturing_quality_0087_open_missing_direct", "split": "train", "domain": "manufacturing_quality", "family": "open_missing_direct", "subset": "core_contrastive", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_release_approval", "may_ship"], "agent_a_complete_predicates": ["has_release_approval", "may_ship"], "agent_b_open_predicates": ["has_supplier_note", "needs_engineer_review", "passed_visual_check"], "facts_positive": ["Part D passed the visual check."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_visual_check": "passed the visual check", "has_release_approval": "has release approval", "has_supplier_note": "has a supplier note", "may_ship": "may ship to customers", "needs_engineer_review": "needs engineer review"}, "vocabulary_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "query_atom": "has_supplier_note::Part D", "target_statement": "Part D has a supplier note.", "gold_reason_type": "local_open_world_underivable", "symbolic": {"positive_atoms": ["passed_visual_check::Part D"], "negative_atoms": [], "rules": [], "query_atom": "has_supplier_note::Part D", "closure_atoms": ["passed_visual_check::Part D"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Part D passed the visual check."], "negative": []}}, "gold_truth_value": "unknown", "gold_source_used": "agent_b", "gold_closure_handling": "open_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_release_approval, may_ship.\nAgent B open/incomplete predicates: has_supplier_note, needs_engineer_review, passed_visual_check.\n\nPredicate glossary:\n- has_release_approval: has release approval\n- has_supplier_note: has a supplier note\n- may_ship: may ship to customers\n- needs_engineer_review: needs engineer review\n- passed_visual_check: passed the visual check\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Part D passed the visual check.\nNegative facts:\n- none\n\nTarget atom: has_supplier_note::Part D.\nTarget statement: Part D has a supplier note.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "manufacturing_quality_0087_open_missing_direct__owa__multi_agent", "source_id": "manufacturing_quality_0087_open_missing_direct__owa", "base_id": "manufacturing_quality_0087_open_missing_direct", "split": "train", "domain": "manufacturing_quality", "family": "open_missing_direct", "subset": "core_contrastive", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "facts_positive": ["Part D passed the visual check."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_visual_check": "passed the visual check", "has_release_approval": "has release approval", "has_supplier_note": "has a supplier note", "may_ship": "may ship to customers", "needs_engineer_review": "needs engineer review"}, "vocabulary_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "query_atom": "has_supplier_note::Part D", "target_statement": "Part D has a supplier note.", "gold_reason_type": "open_world_underivable", "symbolic": {"positive_atoms": ["passed_visual_check::Part D"], "negative_atoms": [], "rules": [], "query_atom": "has_supplier_note::Part D", "closure_atoms": ["passed_visual_check::Part D"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Part D passed the visual check."], "negative": []}}, "gold_truth_value": "unknown", "gold_source_used": "agent_b", "gold_closure_handling": "open_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: has_release_approval, has_supplier_note, may_ship, needs_engineer_review, passed_visual_check.\n\nPredicate glossary:\n- has_release_approval: has release approval\n- has_supplier_note: has a supplier note\n- may_ship: may ship to customers\n- needs_engineer_review: needs engineer review\n- passed_visual_check: passed the visual check\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Part D passed the visual check.\nNegative facts:\n- none\n\nTarget atom: has_supplier_note::Part D.\nTarget statement: Part D has a supplier note.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "manufacturing_quality_0089_closed_missing_with_open_distractor__cwa__multi_agent", "source_id": "manufacturing_quality_0089_closed_missing_with_open_distractor__cwa", "base_id": "manufacturing_quality_0089_closed_missing_with_open_distractor", "split": "dev", "domain": "manufacturing_quality", "family": "closed_missing_with_open_distractor", "subset": "core_contrastive", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "agent_a_complete_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "agent_b_open_predicates": [], "facts_positive": ["Part A has a supplier note.", "Part F passed the visual check."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_visual_check": "passed the visual check", "has_release_approval": "has release approval", "has_supplier_note": "has a supplier note", "may_ship": "may ship to customers", "needs_engineer_review": "needs engineer review"}, "vocabulary_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "query_atom": "has_release_approval::Part F", "target_statement": "Part F has release approval.", "gold_reason_type": "global_closed_world_underivable", "symbolic": {"positive_atoms": ["has_supplier_note::Part A", "passed_visual_check::Part F"], "negative_atoms": [], "rules": [], "query_atom": "has_release_approval::Part F", "closure_atoms": ["has_supplier_note::Part A", "passed_visual_check::Part F"]}, "source_reports": {"agent_a": {"positive": ["Part A has a supplier note.", "Part F passed the visual check."], "negative": []}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "closed_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_release_approval, has_supplier_note, may_ship, needs_engineer_review, passed_visual_check.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- has_release_approval: has release approval\n- has_supplier_note: has a supplier note\n- may_ship: may ship to customers\n- needs_engineer_review: needs engineer review\n- passed_visual_check: passed the visual check\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- Part A has a supplier note.\n- Part F passed the visual check.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: has_release_approval::Part F.\nTarget statement: Part F has release approval.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "manufacturing_quality_0089_closed_missing_with_open_distractor__lcwa__multi_agent", "source_id": "manufacturing_quality_0089_closed_missing_with_open_distractor__lcwa", "base_id": "manufacturing_quality_0089_closed_missing_with_open_distractor", "split": "dev", "domain": "manufacturing_quality", "family": "closed_missing_with_open_distractor", "subset": "core_contrastive", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_release_approval"], "agent_a_complete_predicates": ["has_release_approval"], "agent_b_open_predicates": ["has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "facts_positive": ["Part A has a supplier note.", "Part F passed the visual check."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_visual_check": "passed the visual check", "has_release_approval": "has release approval", "has_supplier_note": "has a supplier note", "may_ship": "may ship to customers", "needs_engineer_review": "needs engineer review"}, "vocabulary_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "query_atom": "has_release_approval::Part F", "target_statement": "Part F has release approval.", "gold_reason_type": "local_closed_world_underivable", "symbolic": {"positive_atoms": ["has_supplier_note::Part A", "passed_visual_check::Part F"], "negative_atoms": [], "rules": [], "query_atom": "has_release_approval::Part F", "closure_atoms": ["has_supplier_note::Part A", "passed_visual_check::Part F"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Part A has a supplier note.", "Part F passed the visual check."], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "closed_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_release_approval.\nAgent B open/incomplete predicates: has_supplier_note, may_ship, needs_engineer_review, passed_visual_check.\n\nPredicate glossary:\n- has_release_approval: has release approval\n- has_supplier_note: has a supplier note\n- may_ship: may ship to customers\n- needs_engineer_review: needs engineer review\n- passed_visual_check: passed the visual check\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Part A has a supplier note.\n- Part F passed the visual check.\nNegative facts:\n- none\n\nTarget atom: has_release_approval::Part F.\nTarget statement: Part F has release approval.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "manufacturing_quality_0089_closed_missing_with_open_distractor__owa__multi_agent", "source_id": "manufacturing_quality_0089_closed_missing_with_open_distractor__owa", "base_id": "manufacturing_quality_0089_closed_missing_with_open_distractor", "split": "dev", "domain": "manufacturing_quality", "family": "closed_missing_with_open_distractor", "subset": "core_contrastive", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "facts_positive": ["Part A has a supplier note.", "Part F passed the visual check."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_visual_check": "passed the visual check", "has_release_approval": "has release approval", "has_supplier_note": "has a supplier note", "may_ship": "may ship to customers", "needs_engineer_review": "needs engineer review"}, "vocabulary_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "query_atom": "has_release_approval::Part F", "target_statement": "Part F has release approval.", "gold_reason_type": "open_world_underivable", "symbolic": {"positive_atoms": ["has_supplier_note::Part A", "passed_visual_check::Part F"], "negative_atoms": [], "rules": [], "query_atom": "has_release_approval::Part F", "closure_atoms": ["has_supplier_note::Part A", "passed_visual_check::Part F"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Part A has a supplier note.", "Part F passed the visual check."], "negative": []}}, "gold_truth_value": "unknown", "gold_source_used": "agent_b", "gold_closure_handling": "open_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: has_release_approval, has_supplier_note, may_ship, needs_engineer_review, passed_visual_check.\n\nPredicate glossary:\n- has_release_approval: has release approval\n- has_supplier_note: has a supplier note\n- may_ship: may ship to customers\n- needs_engineer_review: needs engineer review\n- passed_visual_check: passed the visual check\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Part A has a supplier note.\n- Part F passed the visual check.\nNegative facts:\n- none\n\nTarget atom: has_release_approval::Part F.\nTarget statement: Part F has release approval.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "manufacturing_quality_0090_closed_missing_with_open_distractor__cwa__multi_agent", "source_id": "manufacturing_quality_0090_closed_missing_with_open_distractor__cwa", "base_id": "manufacturing_quality_0090_closed_missing_with_open_distractor", "split": "test", "domain": "manufacturing_quality", "family": "closed_missing_with_open_distractor", "subset": "core_contrastive", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "agent_a_complete_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "agent_b_open_predicates": [], "facts_positive": ["Part B has a supplier note.", "Part A passed the visual check."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_visual_check": "passed the visual check", "has_release_approval": "has release approval", "has_supplier_note": "has a supplier note", "may_ship": "may ship to customers", "needs_engineer_review": "needs engineer review"}, "vocabulary_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "query_atom": "has_release_approval::Part A", "target_statement": "Part A has release approval.", "gold_reason_type": "global_closed_world_underivable", "symbolic": {"positive_atoms": ["has_supplier_note::Part B", "passed_visual_check::Part A"], "negative_atoms": [], "rules": [], "query_atom": "has_release_approval::Part A", "closure_atoms": ["has_supplier_note::Part B", "passed_visual_check::Part A"]}, "source_reports": {"agent_a": {"positive": ["Part B has a supplier note.", "Part A passed the visual check."], "negative": []}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "closed_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_release_approval, has_supplier_note, may_ship, needs_engineer_review, passed_visual_check.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- has_release_approval: has release approval\n- has_supplier_note: has a supplier note\n- may_ship: may ship to customers\n- needs_engineer_review: needs engineer review\n- passed_visual_check: passed the visual check\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- Part B has a supplier note.\n- Part A passed the visual check.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: has_release_approval::Part A.\nTarget statement: Part A has release approval.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "manufacturing_quality_0090_closed_missing_with_open_distractor__lcwa__multi_agent", "source_id": "manufacturing_quality_0090_closed_missing_with_open_distractor__lcwa", "base_id": "manufacturing_quality_0090_closed_missing_with_open_distractor", "split": "test", "domain": "manufacturing_quality", "family": "closed_missing_with_open_distractor", "subset": "core_contrastive", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_release_approval"], "agent_a_complete_predicates": ["has_release_approval"], "agent_b_open_predicates": ["has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "facts_positive": ["Part B has a supplier note.", "Part A passed the visual check."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_visual_check": "passed the visual check", "has_release_approval": "has release approval", "has_supplier_note": "has a supplier note", "may_ship": "may ship to customers", "needs_engineer_review": "needs engineer review"}, "vocabulary_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "query_atom": "has_release_approval::Part A", "target_statement": "Part A has release approval.", "gold_reason_type": "local_closed_world_underivable", "symbolic": {"positive_atoms": ["has_supplier_note::Part B", "passed_visual_check::Part A"], "negative_atoms": [], "rules": [], "query_atom": "has_release_approval::Part A", "closure_atoms": ["has_supplier_note::Part B", "passed_visual_check::Part A"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Part B has a supplier note.", "Part A passed the visual check."], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "closed_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_release_approval.\nAgent B open/incomplete predicates: has_supplier_note, may_ship, needs_engineer_review, passed_visual_check.\n\nPredicate glossary:\n- has_release_approval: has release approval\n- has_supplier_note: has a supplier note\n- may_ship: may ship to customers\n- needs_engineer_review: needs engineer review\n- passed_visual_check: passed the visual check\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Part B has a supplier note.\n- Part A passed the visual check.\nNegative facts:\n- none\n\nTarget atom: has_release_approval::Part A.\nTarget statement: Part A has release approval.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "manufacturing_quality_0090_closed_missing_with_open_distractor__owa__multi_agent", "source_id": "manufacturing_quality_0090_closed_missing_with_open_distractor__owa", "base_id": "manufacturing_quality_0090_closed_missing_with_open_distractor", "split": "test", "domain": "manufacturing_quality", "family": "closed_missing_with_open_distractor", "subset": "core_contrastive", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "facts_positive": ["Part B has a supplier note.", "Part A passed the visual check."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_visual_check": "passed the visual check", "has_release_approval": "has release approval", "has_supplier_note": "has a supplier note", "may_ship": "may ship to customers", "needs_engineer_review": "needs engineer review"}, "vocabulary_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "query_atom": "has_release_approval::Part A", "target_statement": "Part A has release approval.", "gold_reason_type": "open_world_underivable", "symbolic": {"positive_atoms": ["has_supplier_note::Part B", "passed_visual_check::Part A"], "negative_atoms": [], "rules": [], "query_atom": "has_release_approval::Part A", "closure_atoms": ["has_supplier_note::Part B", "passed_visual_check::Part A"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Part B has a supplier note.", "Part A passed the visual check."], "negative": []}}, "gold_truth_value": "unknown", "gold_source_used": "agent_b", "gold_closure_handling": "open_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: has_release_approval, has_supplier_note, may_ship, needs_engineer_review, passed_visual_check.\n\nPredicate glossary:\n- has_release_approval: has release approval\n- has_supplier_note: has a supplier note\n- may_ship: may ship to customers\n- needs_engineer_review: needs engineer review\n- passed_visual_check: passed the visual check\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Part B has a supplier note.\n- Part A passed the visual check.\nNegative facts:\n- none\n\nTarget atom: has_release_approval::Part A.\nTarget statement: Part A has release approval.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "manufacturing_quality_0091_closed_missing_with_open_distractor__cwa__multi_agent", "source_id": "manufacturing_quality_0091_closed_missing_with_open_distractor__cwa", "base_id": "manufacturing_quality_0091_closed_missing_with_open_distractor", "split": "train", "domain": "manufacturing_quality", "family": "closed_missing_with_open_distractor", "subset": "core_contrastive", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "agent_a_complete_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "agent_b_open_predicates": [], "facts_positive": ["Part C has a supplier note.", "Part B passed the visual check."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_visual_check": "passed the visual check", "has_release_approval": "has release approval", "has_supplier_note": "has a supplier note", "may_ship": "may ship to customers", "needs_engineer_review": "needs engineer review"}, "vocabulary_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "query_atom": "has_release_approval::Part B", "target_statement": "Part B has release approval.", "gold_reason_type": "global_closed_world_underivable", "symbolic": {"positive_atoms": ["has_supplier_note::Part C", "passed_visual_check::Part B"], "negative_atoms": [], "rules": [], "query_atom": "has_release_approval::Part B", "closure_atoms": ["has_supplier_note::Part C", "passed_visual_check::Part B"]}, "source_reports": {"agent_a": {"positive": ["Part C has a supplier note.", "Part B passed the visual check."], "negative": []}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "closed_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_release_approval, has_supplier_note, may_ship, needs_engineer_review, passed_visual_check.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- has_release_approval: has release approval\n- has_supplier_note: has a supplier note\n- may_ship: may ship to customers\n- needs_engineer_review: needs engineer review\n- passed_visual_check: passed the visual check\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- Part C has a supplier note.\n- Part B passed the visual check.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: has_release_approval::Part B.\nTarget statement: Part B has release approval.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "manufacturing_quality_0091_closed_missing_with_open_distractor__lcwa__multi_agent", "source_id": "manufacturing_quality_0091_closed_missing_with_open_distractor__lcwa", "base_id": "manufacturing_quality_0091_closed_missing_with_open_distractor", "split": "train", "domain": "manufacturing_quality", "family": "closed_missing_with_open_distractor", "subset": "core_contrastive", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_release_approval"], "agent_a_complete_predicates": ["has_release_approval"], "agent_b_open_predicates": ["has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "facts_positive": ["Part C has a supplier note.", "Part B passed the visual check."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_visual_check": "passed the visual check", "has_release_approval": "has release approval", "has_supplier_note": "has a supplier note", "may_ship": "may ship to customers", "needs_engineer_review": "needs engineer review"}, "vocabulary_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "query_atom": "has_release_approval::Part B", "target_statement": "Part B has release approval.", "gold_reason_type": "local_closed_world_underivable", "symbolic": {"positive_atoms": ["has_supplier_note::Part C", "passed_visual_check::Part B"], "negative_atoms": [], "rules": [], "query_atom": "has_release_approval::Part B", "closure_atoms": ["has_supplier_note::Part C", "passed_visual_check::Part B"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Part C has a supplier note.", "Part B passed the visual check."], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "closed_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_release_approval.\nAgent B open/incomplete predicates: has_supplier_note, may_ship, needs_engineer_review, passed_visual_check.\n\nPredicate glossary:\n- has_release_approval: has release approval\n- has_supplier_note: has a supplier note\n- may_ship: may ship to customers\n- needs_engineer_review: needs engineer review\n- passed_visual_check: passed the visual check\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Part C has a supplier note.\n- Part B passed the visual check.\nNegative facts:\n- none\n\nTarget atom: has_release_approval::Part B.\nTarget statement: Part B has release approval.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "manufacturing_quality_0091_closed_missing_with_open_distractor__owa__multi_agent", "source_id": "manufacturing_quality_0091_closed_missing_with_open_distractor__owa", "base_id": "manufacturing_quality_0091_closed_missing_with_open_distractor", "split": "train", "domain": "manufacturing_quality", "family": "closed_missing_with_open_distractor", "subset": "core_contrastive", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "facts_positive": ["Part C has a supplier note.", "Part B passed the visual check."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_visual_check": "passed the visual check", "has_release_approval": "has release approval", "has_supplier_note": "has a supplier note", "may_ship": "may ship to customers", "needs_engineer_review": "needs engineer review"}, "vocabulary_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "query_atom": "has_release_approval::Part B", "target_statement": "Part B has release approval.", "gold_reason_type": "open_world_underivable", "symbolic": {"positive_atoms": ["has_supplier_note::Part C", "passed_visual_check::Part B"], "negative_atoms": [], "rules": [], "query_atom": "has_release_approval::Part B", "closure_atoms": ["has_supplier_note::Part C", "passed_visual_check::Part B"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Part C has a supplier note.", "Part B passed the visual check."], "negative": []}}, "gold_truth_value": "unknown", "gold_source_used": "agent_b", "gold_closure_handling": "open_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: has_release_approval, has_supplier_note, may_ship, needs_engineer_review, passed_visual_check.\n\nPredicate glossary:\n- has_release_approval: has release approval\n- has_supplier_note: has a supplier note\n- may_ship: may ship to customers\n- needs_engineer_review: needs engineer review\n- passed_visual_check: passed the visual check\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Part C has a supplier note.\n- Part B passed the visual check.\nNegative facts:\n- none\n\nTarget atom: has_release_approval::Part B.\nTarget statement: Part B has release approval.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "manufacturing_quality_0094_open_missing_with_closed_distractor__cwa__multi_agent", "source_id": "manufacturing_quality_0094_open_missing_with_closed_distractor__cwa", "base_id": "manufacturing_quality_0094_open_missing_with_closed_distractor", "split": "test", "domain": "manufacturing_quality", "family": "open_missing_with_closed_distractor", "subset": "core_contrastive", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "agent_a_complete_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "agent_b_open_predicates": [], "facts_positive": ["Part F has release approval.", "Part F passed the visual check."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_visual_check": "passed the visual check", "has_release_approval": "has release approval", "has_supplier_note": "has a supplier note", "may_ship": "may ship to customers", "needs_engineer_review": "needs engineer review"}, "vocabulary_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "query_atom": "has_supplier_note::Part E", "target_statement": "Part E has a supplier note.", "gold_reason_type": "global_closed_world_underivable", "symbolic": {"positive_atoms": ["has_release_approval::Part F", "passed_visual_check::Part F"], "negative_atoms": [], "rules": [], "query_atom": "has_supplier_note::Part E", "closure_atoms": ["has_release_approval::Part F", "passed_visual_check::Part F"]}, "source_reports": {"agent_a": {"positive": ["Part F has release approval.", "Part F passed the visual check."], "negative": []}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "closed_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_release_approval, has_supplier_note, may_ship, needs_engineer_review, passed_visual_check.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- has_release_approval: has release approval\n- has_supplier_note: has a supplier note\n- may_ship: may ship to customers\n- needs_engineer_review: needs engineer review\n- passed_visual_check: passed the visual check\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- Part F has release approval.\n- Part F passed the visual check.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: has_supplier_note::Part E.\nTarget statement: Part E has a supplier note.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "manufacturing_quality_0094_open_missing_with_closed_distractor__lcwa__multi_agent", "source_id": "manufacturing_quality_0094_open_missing_with_closed_distractor__lcwa", "base_id": "manufacturing_quality_0094_open_missing_with_closed_distractor", "split": "test", "domain": "manufacturing_quality", "family": "open_missing_with_closed_distractor", "subset": "core_contrastive", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_release_approval", "may_ship"], "agent_a_complete_predicates": ["has_release_approval", "may_ship"], "agent_b_open_predicates": ["has_supplier_note", "needs_engineer_review", "passed_visual_check"], "facts_positive": ["Part F has release approval.", "Part F passed the visual check."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_visual_check": "passed the visual check", "has_release_approval": "has release approval", "has_supplier_note": "has a supplier note", "may_ship": "may ship to customers", "needs_engineer_review": "needs engineer review"}, "vocabulary_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "query_atom": "has_supplier_note::Part E", "target_statement": "Part E has a supplier note.", "gold_reason_type": "local_open_world_underivable", "symbolic": {"positive_atoms": ["has_release_approval::Part F", "passed_visual_check::Part F"], "negative_atoms": [], "rules": [], "query_atom": "has_supplier_note::Part E", "closure_atoms": ["has_release_approval::Part F", "passed_visual_check::Part F"]}, "source_reports": {"agent_a": {"positive": ["Part F has release approval."], "negative": []}, "agent_b": {"positive": ["Part F passed the visual check."], "negative": []}}, "gold_truth_value": "unknown", "gold_source_used": "agent_b", "gold_closure_handling": "open_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_release_approval, may_ship.\nAgent B open/incomplete predicates: has_supplier_note, needs_engineer_review, passed_visual_check.\n\nPredicate glossary:\n- has_release_approval: has release approval\n- has_supplier_note: has a supplier note\n- may_ship: may ship to customers\n- needs_engineer_review: needs engineer review\n- passed_visual_check: passed the visual check\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- Part F has release approval.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Part F passed the visual check.\nNegative facts:\n- none\n\nTarget atom: has_supplier_note::Part E.\nTarget statement: Part E has a supplier note.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "manufacturing_quality_0094_open_missing_with_closed_distractor__owa__multi_agent", "source_id": "manufacturing_quality_0094_open_missing_with_closed_distractor__owa", "base_id": "manufacturing_quality_0094_open_missing_with_closed_distractor", "split": "test", "domain": "manufacturing_quality", "family": "open_missing_with_closed_distractor", "subset": "core_contrastive", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "facts_positive": ["Part F has release approval.", "Part F passed the visual check."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_visual_check": "passed the visual check", "has_release_approval": "has release approval", "has_supplier_note": "has a supplier note", "may_ship": "may ship to customers", "needs_engineer_review": "needs engineer review"}, "vocabulary_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "query_atom": "has_supplier_note::Part E", "target_statement": "Part E has a supplier note.", "gold_reason_type": "open_world_underivable", "symbolic": {"positive_atoms": ["has_release_approval::Part F", "passed_visual_check::Part F"], "negative_atoms": [], "rules": [], "query_atom": "has_supplier_note::Part E", "closure_atoms": ["has_release_approval::Part F", "passed_visual_check::Part F"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Part F has release approval.", "Part F passed the visual check."], "negative": []}}, "gold_truth_value": "unknown", "gold_source_used": "agent_b", "gold_closure_handling": "open_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: has_release_approval, has_supplier_note, may_ship, needs_engineer_review, passed_visual_check.\n\nPredicate glossary:\n- has_release_approval: has release approval\n- has_supplier_note: has a supplier note\n- may_ship: may ship to customers\n- needs_engineer_review: needs engineer review\n- passed_visual_check: passed the visual check\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Part F has release approval.\n- Part F passed the visual check.\nNegative facts:\n- none\n\nTarget atom: has_supplier_note::Part E.\nTarget statement: Part E has a supplier note.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "manufacturing_quality_0095_open_missing_with_closed_distractor__cwa__multi_agent", "source_id": "manufacturing_quality_0095_open_missing_with_closed_distractor__cwa", "base_id": "manufacturing_quality_0095_open_missing_with_closed_distractor", "split": "train", "domain": "manufacturing_quality", "family": "open_missing_with_closed_distractor", "subset": "core_contrastive", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "agent_a_complete_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "agent_b_open_predicates": [], "facts_positive": ["Part A has release approval.", "Part A passed the visual check."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_visual_check": "passed the visual check", "has_release_approval": "has release approval", "has_supplier_note": "has a supplier note", "may_ship": "may ship to customers", "needs_engineer_review": "needs engineer review"}, "vocabulary_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "query_atom": "has_supplier_note::Part F", "target_statement": "Part F has a supplier note.", "gold_reason_type": "global_closed_world_underivable", "symbolic": {"positive_atoms": ["has_release_approval::Part A", "passed_visual_check::Part A"], "negative_atoms": [], "rules": [], "query_atom": "has_supplier_note::Part F", "closure_atoms": ["has_release_approval::Part A", "passed_visual_check::Part A"]}, "source_reports": {"agent_a": {"positive": ["Part A has release approval.", "Part A passed the visual check."], "negative": []}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "closed_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_release_approval, has_supplier_note, may_ship, needs_engineer_review, passed_visual_check.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- has_release_approval: has release approval\n- has_supplier_note: has a supplier note\n- may_ship: may ship to customers\n- needs_engineer_review: needs engineer review\n- passed_visual_check: passed the visual check\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- Part A has release approval.\n- Part A passed the visual check.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: has_supplier_note::Part F.\nTarget statement: Part F has a supplier note.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "manufacturing_quality_0095_open_missing_with_closed_distractor__lcwa__multi_agent", "source_id": "manufacturing_quality_0095_open_missing_with_closed_distractor__lcwa", "base_id": "manufacturing_quality_0095_open_missing_with_closed_distractor", "split": "train", "domain": "manufacturing_quality", "family": "open_missing_with_closed_distractor", "subset": "core_contrastive", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_release_approval", "may_ship"], "agent_a_complete_predicates": ["has_release_approval", "may_ship"], "agent_b_open_predicates": ["has_supplier_note", "needs_engineer_review", "passed_visual_check"], "facts_positive": ["Part A has release approval.", "Part A passed the visual check."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_visual_check": "passed the visual check", "has_release_approval": "has release approval", "has_supplier_note": "has a supplier note", "may_ship": "may ship to customers", "needs_engineer_review": "needs engineer review"}, "vocabulary_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "query_atom": "has_supplier_note::Part F", "target_statement": "Part F has a supplier note.", "gold_reason_type": "local_open_world_underivable", "symbolic": {"positive_atoms": ["has_release_approval::Part A", "passed_visual_check::Part A"], "negative_atoms": [], "rules": [], "query_atom": "has_supplier_note::Part F", "closure_atoms": ["has_release_approval::Part A", "passed_visual_check::Part A"]}, "source_reports": {"agent_a": {"positive": ["Part A has release approval."], "negative": []}, "agent_b": {"positive": ["Part A passed the visual check."], "negative": []}}, "gold_truth_value": "unknown", "gold_source_used": "agent_b", "gold_closure_handling": "open_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_release_approval, may_ship.\nAgent B open/incomplete predicates: has_supplier_note, needs_engineer_review, passed_visual_check.\n\nPredicate glossary:\n- has_release_approval: has release approval\n- has_supplier_note: has a supplier note\n- may_ship: may ship to customers\n- needs_engineer_review: needs engineer review\n- passed_visual_check: passed the visual check\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- Part A has release approval.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Part A passed the visual check.\nNegative facts:\n- none\n\nTarget atom: has_supplier_note::Part F.\nTarget statement: Part F has a supplier note.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "manufacturing_quality_0095_open_missing_with_closed_distractor__owa__multi_agent", "source_id": "manufacturing_quality_0095_open_missing_with_closed_distractor__owa", "base_id": "manufacturing_quality_0095_open_missing_with_closed_distractor", "split": "train", "domain": "manufacturing_quality", "family": "open_missing_with_closed_distractor", "subset": "core_contrastive", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "facts_positive": ["Part A has release approval.", "Part A passed the visual check."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_visual_check": "passed the visual check", "has_release_approval": "has release approval", "has_supplier_note": "has a supplier note", "may_ship": "may ship to customers", "needs_engineer_review": "needs engineer review"}, "vocabulary_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "query_atom": "has_supplier_note::Part F", "target_statement": "Part F has a supplier note.", "gold_reason_type": "open_world_underivable", "symbolic": {"positive_atoms": ["has_release_approval::Part A", "passed_visual_check::Part A"], "negative_atoms": [], "rules": [], "query_atom": "has_supplier_note::Part F", "closure_atoms": ["has_release_approval::Part A", "passed_visual_check::Part A"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Part A has release approval.", "Part A passed the visual check."], "negative": []}}, "gold_truth_value": "unknown", "gold_source_used": "agent_b", "gold_closure_handling": "open_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: has_release_approval, has_supplier_note, may_ship, needs_engineer_review, passed_visual_check.\n\nPredicate glossary:\n- has_release_approval: has release approval\n- has_supplier_note: has a supplier note\n- may_ship: may ship to customers\n- needs_engineer_review: needs engineer review\n- passed_visual_check: passed the visual check\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Part A has release approval.\n- Part A passed the visual check.\nNegative facts:\n- none\n\nTarget atom: has_supplier_note::Part F.\nTarget statement: Part F has a supplier note.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "manufacturing_quality_0096_closed_derived_missing_antecedent__cwa__multi_agent", "source_id": "manufacturing_quality_0096_closed_derived_missing_antecedent__cwa", "base_id": "manufacturing_quality_0096_closed_derived_missing_antecedent", "split": "train", "domain": "manufacturing_quality", "family": "closed_derived_missing_antecedent", "subset": "core_contrastive", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "agent_a_complete_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "agent_b_open_predicates": [], "facts_positive": ["Part A passed the visual check."], "facts_negative": [], "rules_natural": ["If a part passed the visual check and has release approval, then that part may ship to customers."], "predicate_glossary": {"passed_visual_check": "passed the visual check", "has_release_approval": "has release approval", "has_supplier_note": "has a supplier note", "may_ship": "may ship to customers", "needs_engineer_review": "needs engineer review"}, "vocabulary_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "query_atom": "may_ship::Part A", "target_statement": "Part A may ship to customers.", "gold_reason_type": "global_closed_world_underivable", "symbolic": {"positive_atoms": ["passed_visual_check::Part A"], "negative_atoms": [], "rules": [{"antecedents": ["passed_visual_check::Part A", "has_release_approval::Part A"], "conclusion": "may_ship::Part A", "text": "If a part passed the visual check and has release approval, then that part may ship to customers."}], "query_atom": "may_ship::Part A", "closure_atoms": ["passed_visual_check::Part A"]}, "source_reports": {"agent_a": {"positive": ["Part A passed the visual check."], "negative": []}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "closed_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_release_approval, has_supplier_note, may_ship, needs_engineer_review, passed_visual_check.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- has_release_approval: has release approval\n- has_supplier_note: has a supplier note\n- may_ship: may ship to customers\n- needs_engineer_review: needs engineer review\n- passed_visual_check: passed the visual check\n\nCoordinator policy rules:\n- If a part passed the visual check and has release approval, then that part may ship to customers.\n\nAgent A report:\nPositive facts:\n- Part A passed the visual check.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: may_ship::Part A.\nTarget statement: Part A may ship to customers.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "manufacturing_quality_0096_closed_derived_missing_antecedent__lcwa__multi_agent", "source_id": "manufacturing_quality_0096_closed_derived_missing_antecedent__lcwa", "base_id": "manufacturing_quality_0096_closed_derived_missing_antecedent", "split": "train", "domain": "manufacturing_quality", "family": "closed_derived_missing_antecedent", "subset": "core_contrastive", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_release_approval", "may_ship"], "agent_a_complete_predicates": ["has_release_approval", "may_ship"], "agent_b_open_predicates": ["has_supplier_note", "needs_engineer_review", "passed_visual_check"], "facts_positive": ["Part A passed the visual check."], "facts_negative": [], "rules_natural": ["If a part passed the visual check and has release approval, then that part may ship to customers."], "predicate_glossary": {"passed_visual_check": "passed the visual check", "has_release_approval": "has release approval", "has_supplier_note": "has a supplier note", "may_ship": "may ship to customers", "needs_engineer_review": "needs engineer review"}, "vocabulary_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "query_atom": "may_ship::Part A", "target_statement": "Part A may ship to customers.", "gold_reason_type": "local_closed_world_underivable", "symbolic": {"positive_atoms": ["passed_visual_check::Part A"], "negative_atoms": [], "rules": [{"antecedents": ["passed_visual_check::Part A", "has_release_approval::Part A"], "conclusion": "may_ship::Part A", "text": "If a part passed the visual check and has release approval, then that part may ship to customers."}], "query_atom": "may_ship::Part A", "closure_atoms": ["passed_visual_check::Part A"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Part A passed the visual check."], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "closed_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_release_approval, may_ship.\nAgent B open/incomplete predicates: has_supplier_note, needs_engineer_review, passed_visual_check.\n\nPredicate glossary:\n- has_release_approval: has release approval\n- has_supplier_note: has a supplier note\n- may_ship: may ship to customers\n- needs_engineer_review: needs engineer review\n- passed_visual_check: passed the visual check\n\nCoordinator policy rules:\n- If a part passed the visual check and has release approval, then that part may ship to customers.\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Part A passed the visual check.\nNegative facts:\n- none\n\nTarget atom: may_ship::Part A.\nTarget statement: Part A may ship to customers.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "manufacturing_quality_0096_closed_derived_missing_antecedent__owa__multi_agent", "source_id": "manufacturing_quality_0096_closed_derived_missing_antecedent__owa", "base_id": "manufacturing_quality_0096_closed_derived_missing_antecedent", "split": "train", "domain": "manufacturing_quality", "family": "closed_derived_missing_antecedent", "subset": "core_contrastive", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "facts_positive": ["Part A passed the visual check."], "facts_negative": [], "rules_natural": ["If a part passed the visual check and has release approval, then that part may ship to customers."], "predicate_glossary": {"passed_visual_check": "passed the visual check", "has_release_approval": "has release approval", "has_supplier_note": "has a supplier note", "may_ship": "may ship to customers", "needs_engineer_review": "needs engineer review"}, "vocabulary_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "query_atom": "may_ship::Part A", "target_statement": "Part A may ship to customers.", "gold_reason_type": "open_world_underivable", "symbolic": {"positive_atoms": ["passed_visual_check::Part A"], "negative_atoms": [], "rules": [{"antecedents": ["passed_visual_check::Part A", "has_release_approval::Part A"], "conclusion": "may_ship::Part A", "text": "If a part passed the visual check and has release approval, then that part may ship to customers."}], "query_atom": "may_ship::Part A", "closure_atoms": ["passed_visual_check::Part A"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Part A passed the visual check."], "negative": []}}, "gold_truth_value": "unknown", "gold_source_used": "agent_b", "gold_closure_handling": "open_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: has_release_approval, has_supplier_note, may_ship, needs_engineer_review, passed_visual_check.\n\nPredicate glossary:\n- has_release_approval: has release approval\n- has_supplier_note: has a supplier note\n- may_ship: may ship to customers\n- needs_engineer_review: needs engineer review\n- passed_visual_check: passed the visual check\n\nCoordinator policy rules:\n- If a part passed the visual check and has release approval, then that part may ship to customers.\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Part A passed the visual check.\nNegative facts:\n- none\n\nTarget atom: may_ship::Part A.\nTarget statement: Part A may ship to customers.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "manufacturing_quality_0098_closed_derived_missing_antecedent__cwa__multi_agent", "source_id": "manufacturing_quality_0098_closed_derived_missing_antecedent__cwa", "base_id": "manufacturing_quality_0098_closed_derived_missing_antecedent", "split": "test", "domain": "manufacturing_quality", "family": "closed_derived_missing_antecedent", "subset": "core_contrastive", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "agent_a_complete_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "agent_b_open_predicates": [], "facts_positive": ["Part C passed the visual check."], "facts_negative": [], "rules_natural": ["If a part passed the visual check and has release approval, then that part may ship to customers."], "predicate_glossary": {"passed_visual_check": "passed the visual check", "has_release_approval": "has release approval", "has_supplier_note": "has a supplier note", "may_ship": "may ship to customers", "needs_engineer_review": "needs engineer review"}, "vocabulary_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "query_atom": "may_ship::Part C", "target_statement": "Part C may ship to customers.", "gold_reason_type": "global_closed_world_underivable", "symbolic": {"positive_atoms": ["passed_visual_check::Part C"], "negative_atoms": [], "rules": [{"antecedents": ["passed_visual_check::Part C", "has_release_approval::Part C"], "conclusion": "may_ship::Part C", "text": "If a part passed the visual check and has release approval, then that part may ship to customers."}], "query_atom": "may_ship::Part C", "closure_atoms": ["passed_visual_check::Part C"]}, "source_reports": {"agent_a": {"positive": ["Part C passed the visual check."], "negative": []}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "closed_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_release_approval, has_supplier_note, may_ship, needs_engineer_review, passed_visual_check.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- has_release_approval: has release approval\n- has_supplier_note: has a supplier note\n- may_ship: may ship to customers\n- needs_engineer_review: needs engineer review\n- passed_visual_check: passed the visual check\n\nCoordinator policy rules:\n- If a part passed the visual check and has release approval, then that part may ship to customers.\n\nAgent A report:\nPositive facts:\n- Part C passed the visual check.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: may_ship::Part C.\nTarget statement: Part C may ship to customers.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "manufacturing_quality_0098_closed_derived_missing_antecedent__lcwa__multi_agent", "source_id": "manufacturing_quality_0098_closed_derived_missing_antecedent__lcwa", "base_id": "manufacturing_quality_0098_closed_derived_missing_antecedent", "split": "test", "domain": "manufacturing_quality", "family": "closed_derived_missing_antecedent", "subset": "core_contrastive", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_release_approval", "may_ship"], "agent_a_complete_predicates": ["has_release_approval", "may_ship"], "agent_b_open_predicates": ["has_supplier_note", "needs_engineer_review", "passed_visual_check"], "facts_positive": ["Part C passed the visual check."], "facts_negative": [], "rules_natural": ["If a part passed the visual check and has release approval, then that part may ship to customers."], "predicate_glossary": {"passed_visual_check": "passed the visual check", "has_release_approval": "has release approval", "has_supplier_note": "has a supplier note", "may_ship": "may ship to customers", "needs_engineer_review": "needs engineer review"}, "vocabulary_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "query_atom": "may_ship::Part C", "target_statement": "Part C may ship to customers.", "gold_reason_type": "local_closed_world_underivable", "symbolic": {"positive_atoms": ["passed_visual_check::Part C"], "negative_atoms": [], "rules": [{"antecedents": ["passed_visual_check::Part C", "has_release_approval::Part C"], "conclusion": "may_ship::Part C", "text": "If a part passed the visual check and has release approval, then that part may ship to customers."}], "query_atom": "may_ship::Part C", "closure_atoms": ["passed_visual_check::Part C"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Part C passed the visual check."], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "closed_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_release_approval, may_ship.\nAgent B open/incomplete predicates: has_supplier_note, needs_engineer_review, passed_visual_check.\n\nPredicate glossary:\n- has_release_approval: has release approval\n- has_supplier_note: has a supplier note\n- may_ship: may ship to customers\n- needs_engineer_review: needs engineer review\n- passed_visual_check: passed the visual check\n\nCoordinator policy rules:\n- If a part passed the visual check and has release approval, then that part may ship to customers.\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Part C passed the visual check.\nNegative facts:\n- none\n\nTarget atom: may_ship::Part C.\nTarget statement: Part C may ship to customers.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "manufacturing_quality_0098_closed_derived_missing_antecedent__owa__multi_agent", "source_id": "manufacturing_quality_0098_closed_derived_missing_antecedent__owa", "base_id": "manufacturing_quality_0098_closed_derived_missing_antecedent", "split": "test", "domain": "manufacturing_quality", "family": "closed_derived_missing_antecedent", "subset": "core_contrastive", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "facts_positive": ["Part C passed the visual check."], "facts_negative": [], "rules_natural": ["If a part passed the visual check and has release approval, then that part may ship to customers."], "predicate_glossary": {"passed_visual_check": "passed the visual check", "has_release_approval": "has release approval", "has_supplier_note": "has a supplier note", "may_ship": "may ship to customers", "needs_engineer_review": "needs engineer review"}, "vocabulary_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "query_atom": "may_ship::Part C", "target_statement": "Part C may ship to customers.", "gold_reason_type": "open_world_underivable", "symbolic": {"positive_atoms": ["passed_visual_check::Part C"], "negative_atoms": [], "rules": [{"antecedents": ["passed_visual_check::Part C", "has_release_approval::Part C"], "conclusion": "may_ship::Part C", "text": "If a part passed the visual check and has release approval, then that part may ship to customers."}], "query_atom": "may_ship::Part C", "closure_atoms": ["passed_visual_check::Part C"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Part C passed the visual check."], "negative": []}}, "gold_truth_value": "unknown", "gold_source_used": "agent_b", "gold_closure_handling": "open_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: has_release_approval, has_supplier_note, may_ship, needs_engineer_review, passed_visual_check.\n\nPredicate glossary:\n- has_release_approval: has release approval\n- has_supplier_note: has a supplier note\n- may_ship: may ship to customers\n- needs_engineer_review: needs engineer review\n- passed_visual_check: passed the visual check\n\nCoordinator policy rules:\n- If a part passed the visual check and has release approval, then that part may ship to customers.\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Part C passed the visual check.\nNegative facts:\n- none\n\nTarget atom: may_ship::Part C.\nTarget statement: Part C may ship to customers.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "manufacturing_quality_0100_open_derived_missing_antecedent__cwa__multi_agent", "source_id": "manufacturing_quality_0100_open_derived_missing_antecedent__cwa", "base_id": "manufacturing_quality_0100_open_derived_missing_antecedent", "split": "train", "domain": "manufacturing_quality", "family": "open_derived_missing_antecedent", "subset": "core_contrastive", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "agent_a_complete_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "agent_b_open_predicates": [], "facts_positive": ["Part E passed the visual check."], "facts_negative": [], "rules_natural": ["If a part has a supplier note, then that part needs engineer review."], "predicate_glossary": {"passed_visual_check": "passed the visual check", "has_release_approval": "has release approval", "has_supplier_note": "has a supplier note", "may_ship": "may ship to customers", "needs_engineer_review": "needs engineer review"}, "vocabulary_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "query_atom": "needs_engineer_review::Part E", "target_statement": "Part E needs engineer review.", "gold_reason_type": "global_closed_world_underivable", "symbolic": {"positive_atoms": ["passed_visual_check::Part E"], "negative_atoms": [], "rules": [{"antecedents": ["has_supplier_note::Part E"], "conclusion": "needs_engineer_review::Part E", "text": "If a part has a supplier note, then that part needs engineer review."}], "query_atom": "needs_engineer_review::Part E", "closure_atoms": ["passed_visual_check::Part E"]}, "source_reports": {"agent_a": {"positive": ["Part E passed the visual check."], "negative": []}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "closed_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_release_approval, has_supplier_note, may_ship, needs_engineer_review, passed_visual_check.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- has_release_approval: has release approval\n- has_supplier_note: has a supplier note\n- may_ship: may ship to customers\n- needs_engineer_review: needs engineer review\n- passed_visual_check: passed the visual check\n\nCoordinator policy rules:\n- If a part has a supplier note, then that part needs engineer review.\n\nAgent A report:\nPositive facts:\n- Part E passed the visual check.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: needs_engineer_review::Part E.\nTarget statement: Part E needs engineer review.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "manufacturing_quality_0100_open_derived_missing_antecedent__lcwa__multi_agent", "source_id": "manufacturing_quality_0100_open_derived_missing_antecedent__lcwa", "base_id": "manufacturing_quality_0100_open_derived_missing_antecedent", "split": "train", "domain": "manufacturing_quality", "family": "open_derived_missing_antecedent", "subset": "core_contrastive", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_release_approval", "may_ship"], "agent_a_complete_predicates": ["has_release_approval", "may_ship"], "agent_b_open_predicates": ["has_supplier_note", "needs_engineer_review", "passed_visual_check"], "facts_positive": ["Part E passed the visual check."], "facts_negative": [], "rules_natural": ["If a part has a supplier note, then that part needs engineer review."], "predicate_glossary": {"passed_visual_check": "passed the visual check", "has_release_approval": "has release approval", "has_supplier_note": "has a supplier note", "may_ship": "may ship to customers", "needs_engineer_review": "needs engineer review"}, "vocabulary_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "query_atom": "needs_engineer_review::Part E", "target_statement": "Part E needs engineer review.", "gold_reason_type": "local_open_world_underivable", "symbolic": {"positive_atoms": ["passed_visual_check::Part E"], "negative_atoms": [], "rules": [{"antecedents": ["has_supplier_note::Part E"], "conclusion": "needs_engineer_review::Part E", "text": "If a part has a supplier note, then that part needs engineer review."}], "query_atom": "needs_engineer_review::Part E", "closure_atoms": ["passed_visual_check::Part E"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Part E passed the visual check."], "negative": []}}, "gold_truth_value": "unknown", "gold_source_used": "agent_b", "gold_closure_handling": "open_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_release_approval, may_ship.\nAgent B open/incomplete predicates: has_supplier_note, needs_engineer_review, passed_visual_check.\n\nPredicate glossary:\n- has_release_approval: has release approval\n- has_supplier_note: has a supplier note\n- may_ship: may ship to customers\n- needs_engineer_review: needs engineer review\n- passed_visual_check: passed the visual check\n\nCoordinator policy rules:\n- If a part has a supplier note, then that part needs engineer review.\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Part E passed the visual check.\nNegative facts:\n- none\n\nTarget atom: needs_engineer_review::Part E.\nTarget statement: Part E needs engineer review.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "manufacturing_quality_0100_open_derived_missing_antecedent__owa__multi_agent", "source_id": "manufacturing_quality_0100_open_derived_missing_antecedent__owa", "base_id": "manufacturing_quality_0100_open_derived_missing_antecedent", "split": "train", "domain": "manufacturing_quality", "family": "open_derived_missing_antecedent", "subset": "core_contrastive", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "facts_positive": ["Part E passed the visual check."], "facts_negative": [], "rules_natural": ["If a part has a supplier note, then that part needs engineer review."], "predicate_glossary": {"passed_visual_check": "passed the visual check", "has_release_approval": "has release approval", "has_supplier_note": "has a supplier note", "may_ship": "may ship to customers", "needs_engineer_review": "needs engineer review"}, "vocabulary_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "query_atom": "needs_engineer_review::Part E", "target_statement": "Part E needs engineer review.", "gold_reason_type": "open_world_underivable", "symbolic": {"positive_atoms": ["passed_visual_check::Part E"], "negative_atoms": [], "rules": [{"antecedents": ["has_supplier_note::Part E"], "conclusion": "needs_engineer_review::Part E", "text": "If a part has a supplier note, then that part needs engineer review."}], "query_atom": "needs_engineer_review::Part E", "closure_atoms": ["passed_visual_check::Part E"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Part E passed the visual check."], "negative": []}}, "gold_truth_value": "unknown", "gold_source_used": "agent_b", "gold_closure_handling": "open_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: has_release_approval, has_supplier_note, may_ship, needs_engineer_review, passed_visual_check.\n\nPredicate glossary:\n- has_release_approval: has release approval\n- has_supplier_note: has a supplier note\n- may_ship: may ship to customers\n- needs_engineer_review: needs engineer review\n- passed_visual_check: passed the visual check\n\nCoordinator policy rules:\n- If a part has a supplier note, then that part needs engineer review.\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Part E passed the visual check.\nNegative facts:\n- none\n\nTarget atom: needs_engineer_review::Part E.\nTarget statement: Part E needs engineer review.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "manufacturing_quality_0105_entailed_closed_conclusion__cwa__multi_agent", "source_id": "manufacturing_quality_0105_entailed_closed_conclusion__cwa", "base_id": "manufacturing_quality_0105_entailed_closed_conclusion", "split": "test", "domain": "manufacturing_quality", "family": "entailed_closed_conclusion", "subset": "control_entailed", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "agent_a_complete_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "agent_b_open_predicates": [], "facts_positive": ["Part D has release approval.", "Part D passed the visual check."], "facts_negative": [], "rules_natural": ["If a part passed the visual check and has release approval, then that part may ship to customers."], "predicate_glossary": {"passed_visual_check": "passed the visual check", "has_release_approval": "has release approval", "has_supplier_note": "has a supplier note", "may_ship": "may ship to customers", "needs_engineer_review": "needs engineer review"}, "vocabulary_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "query_atom": "may_ship::Part D", "target_statement": "Part D may ship to customers.", "gold_reason_type": "entailed_by_fact_or_rule", "symbolic": {"positive_atoms": ["has_release_approval::Part D", "passed_visual_check::Part D"], "negative_atoms": [], "rules": [{"antecedents": ["passed_visual_check::Part D", "has_release_approval::Part D"], "conclusion": "may_ship::Part D", "text": "If a part passed the visual check and has release approval, then that part may ship to customers."}], "query_atom": "may_ship::Part D", "closure_atoms": ["has_release_approval::Part D", "may_ship::Part D", "passed_visual_check::Part D"]}, "source_reports": {"agent_a": {"positive": ["Part D has release approval.", "Part D passed the visual check."], "negative": []}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "true", "gold_source_used": "agent_a", "gold_closure_handling": "entailed_or_explicit", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_release_approval, has_supplier_note, may_ship, needs_engineer_review, passed_visual_check.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- has_release_approval: has release approval\n- has_supplier_note: has a supplier note\n- may_ship: may ship to customers\n- needs_engineer_review: needs engineer review\n- passed_visual_check: passed the visual check\n\nCoordinator policy rules:\n- If a part passed the visual check and has release approval, then that part may ship to customers.\n\nAgent A report:\nPositive facts:\n- Part D has release approval.\n- Part D passed the visual check.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: may_ship::Part D.\nTarget statement: Part D may ship to customers.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "manufacturing_quality_0105_entailed_closed_conclusion__lcwa__multi_agent", "source_id": "manufacturing_quality_0105_entailed_closed_conclusion__lcwa", "base_id": "manufacturing_quality_0105_entailed_closed_conclusion", "split": "test", "domain": "manufacturing_quality", "family": "entailed_closed_conclusion", "subset": "control_entailed", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_release_approval", "may_ship"], "agent_a_complete_predicates": ["has_release_approval", "may_ship"], "agent_b_open_predicates": ["has_supplier_note", "needs_engineer_review", "passed_visual_check"], "facts_positive": ["Part D has release approval.", "Part D passed the visual check."], "facts_negative": [], "rules_natural": ["If a part passed the visual check and has release approval, then that part may ship to customers."], "predicate_glossary": {"passed_visual_check": "passed the visual check", "has_release_approval": "has release approval", "has_supplier_note": "has a supplier note", "may_ship": "may ship to customers", "needs_engineer_review": "needs engineer review"}, "vocabulary_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "query_atom": "may_ship::Part D", "target_statement": "Part D may ship to customers.", "gold_reason_type": "entailed_by_fact_or_rule", "symbolic": {"positive_atoms": ["has_release_approval::Part D", "passed_visual_check::Part D"], "negative_atoms": [], "rules": [{"antecedents": ["passed_visual_check::Part D", "has_release_approval::Part D"], "conclusion": "may_ship::Part D", "text": "If a part passed the visual check and has release approval, then that part may ship to customers."}], "query_atom": "may_ship::Part D", "closure_atoms": ["has_release_approval::Part D", "may_ship::Part D", "passed_visual_check::Part D"]}, "source_reports": {"agent_a": {"positive": ["Part D has release approval."], "negative": []}, "agent_b": {"positive": ["Part D passed the visual check."], "negative": []}}, "gold_truth_value": "true", "gold_source_used": "both", "gold_closure_handling": "entailed_or_explicit", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_release_approval, may_ship.\nAgent B open/incomplete predicates: has_supplier_note, needs_engineer_review, passed_visual_check.\n\nPredicate glossary:\n- has_release_approval: has release approval\n- has_supplier_note: has a supplier note\n- may_ship: may ship to customers\n- needs_engineer_review: needs engineer review\n- passed_visual_check: passed the visual check\n\nCoordinator policy rules:\n- If a part passed the visual check and has release approval, then that part may ship to customers.\n\nAgent A report:\nPositive facts:\n- Part D has release approval.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Part D passed the visual check.\nNegative facts:\n- none\n\nTarget atom: may_ship::Part D.\nTarget statement: Part D may ship to customers.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "manufacturing_quality_0105_entailed_closed_conclusion__owa__multi_agent", "source_id": "manufacturing_quality_0105_entailed_closed_conclusion__owa", "base_id": "manufacturing_quality_0105_entailed_closed_conclusion", "split": "test", "domain": "manufacturing_quality", "family": "entailed_closed_conclusion", "subset": "control_entailed", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "facts_positive": ["Part D has release approval.", "Part D passed the visual check."], "facts_negative": [], "rules_natural": ["If a part passed the visual check and has release approval, then that part may ship to customers."], "predicate_glossary": {"passed_visual_check": "passed the visual check", "has_release_approval": "has release approval", "has_supplier_note": "has a supplier note", "may_ship": "may ship to customers", "needs_engineer_review": "needs engineer review"}, "vocabulary_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "query_atom": "may_ship::Part D", "target_statement": "Part D may ship to customers.", "gold_reason_type": "entailed_by_fact_or_rule", "symbolic": {"positive_atoms": ["has_release_approval::Part D", "passed_visual_check::Part D"], "negative_atoms": [], "rules": [{"antecedents": ["passed_visual_check::Part D", "has_release_approval::Part D"], "conclusion": "may_ship::Part D", "text": "If a part passed the visual check and has release approval, then that part may ship to customers."}], "query_atom": "may_ship::Part D", "closure_atoms": ["has_release_approval::Part D", "may_ship::Part D", "passed_visual_check::Part D"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Part D has release approval.", "Part D passed the visual check."], "negative": []}}, "gold_truth_value": "true", "gold_source_used": "agent_b", "gold_closure_handling": "entailed_or_explicit", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: has_release_approval, has_supplier_note, may_ship, needs_engineer_review, passed_visual_check.\n\nPredicate glossary:\n- has_release_approval: has release approval\n- has_supplier_note: has a supplier note\n- may_ship: may ship to customers\n- needs_engineer_review: needs engineer review\n- passed_visual_check: passed the visual check\n\nCoordinator policy rules:\n- If a part passed the visual check and has release approval, then that part may ship to customers.\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Part D has release approval.\n- Part D passed the visual check.\nNegative facts:\n- none\n\nTarget atom: may_ship::Part D.\nTarget statement: Part D may ship to customers.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "manufacturing_quality_0106_entailed_closed_conclusion__cwa__multi_agent", "source_id": "manufacturing_quality_0106_entailed_closed_conclusion__cwa", "base_id": "manufacturing_quality_0106_entailed_closed_conclusion", "split": "train", "domain": "manufacturing_quality", "family": "entailed_closed_conclusion", "subset": "control_entailed", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "agent_a_complete_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "agent_b_open_predicates": [], "facts_positive": ["Part E has release approval.", "Part E passed the visual check."], "facts_negative": [], "rules_natural": ["If a part passed the visual check and has release approval, then that part may ship to customers."], "predicate_glossary": {"passed_visual_check": "passed the visual check", "has_release_approval": "has release approval", "has_supplier_note": "has a supplier note", "may_ship": "may ship to customers", "needs_engineer_review": "needs engineer review"}, "vocabulary_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "query_atom": "may_ship::Part E", "target_statement": "Part E may ship to customers.", "gold_reason_type": "entailed_by_fact_or_rule", "symbolic": {"positive_atoms": ["has_release_approval::Part E", "passed_visual_check::Part E"], "negative_atoms": [], "rules": [{"antecedents": ["passed_visual_check::Part E", "has_release_approval::Part E"], "conclusion": "may_ship::Part E", "text": "If a part passed the visual check and has release approval, then that part may ship to customers."}], "query_atom": "may_ship::Part E", "closure_atoms": ["has_release_approval::Part E", "may_ship::Part E", "passed_visual_check::Part E"]}, "source_reports": {"agent_a": {"positive": ["Part E has release approval.", "Part E passed the visual check."], "negative": []}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "true", "gold_source_used": "agent_a", "gold_closure_handling": "entailed_or_explicit", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_release_approval, has_supplier_note, may_ship, needs_engineer_review, passed_visual_check.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- has_release_approval: has release approval\n- has_supplier_note: has a supplier note\n- may_ship: may ship to customers\n- needs_engineer_review: needs engineer review\n- passed_visual_check: passed the visual check\n\nCoordinator policy rules:\n- If a part passed the visual check and has release approval, then that part may ship to customers.\n\nAgent A report:\nPositive facts:\n- Part E has release approval.\n- Part E passed the visual check.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: may_ship::Part E.\nTarget statement: Part E may ship to customers.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "manufacturing_quality_0106_entailed_closed_conclusion__lcwa__multi_agent", "source_id": "manufacturing_quality_0106_entailed_closed_conclusion__lcwa", "base_id": "manufacturing_quality_0106_entailed_closed_conclusion", "split": "train", "domain": "manufacturing_quality", "family": "entailed_closed_conclusion", "subset": "control_entailed", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_release_approval", "may_ship"], "agent_a_complete_predicates": ["has_release_approval", "may_ship"], "agent_b_open_predicates": ["has_supplier_note", "needs_engineer_review", "passed_visual_check"], "facts_positive": ["Part E has release approval.", "Part E passed the visual check."], "facts_negative": [], "rules_natural": ["If a part passed the visual check and has release approval, then that part may ship to customers."], "predicate_glossary": {"passed_visual_check": "passed the visual check", "has_release_approval": "has release approval", "has_supplier_note": "has a supplier note", "may_ship": "may ship to customers", "needs_engineer_review": "needs engineer review"}, "vocabulary_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "query_atom": "may_ship::Part E", "target_statement": "Part E may ship to customers.", "gold_reason_type": "entailed_by_fact_or_rule", "symbolic": {"positive_atoms": ["has_release_approval::Part E", "passed_visual_check::Part E"], "negative_atoms": [], "rules": [{"antecedents": ["passed_visual_check::Part E", "has_release_approval::Part E"], "conclusion": "may_ship::Part E", "text": "If a part passed the visual check and has release approval, then that part may ship to customers."}], "query_atom": "may_ship::Part E", "closure_atoms": ["has_release_approval::Part E", "may_ship::Part E", "passed_visual_check::Part E"]}, "source_reports": {"agent_a": {"positive": ["Part E has release approval."], "negative": []}, "agent_b": {"positive": ["Part E passed the visual check."], "negative": []}}, "gold_truth_value": "true", "gold_source_used": "both", "gold_closure_handling": "entailed_or_explicit", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_release_approval, may_ship.\nAgent B open/incomplete predicates: has_supplier_note, needs_engineer_review, passed_visual_check.\n\nPredicate glossary:\n- has_release_approval: has release approval\n- has_supplier_note: has a supplier note\n- may_ship: may ship to customers\n- needs_engineer_review: needs engineer review\n- passed_visual_check: passed the visual check\n\nCoordinator policy rules:\n- If a part passed the visual check and has release approval, then that part may ship to customers.\n\nAgent A report:\nPositive facts:\n- Part E has release approval.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Part E passed the visual check.\nNegative facts:\n- none\n\nTarget atom: may_ship::Part E.\nTarget statement: Part E may ship to customers.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "manufacturing_quality_0106_entailed_closed_conclusion__owa__multi_agent", "source_id": "manufacturing_quality_0106_entailed_closed_conclusion__owa", "base_id": "manufacturing_quality_0106_entailed_closed_conclusion", "split": "train", "domain": "manufacturing_quality", "family": "entailed_closed_conclusion", "subset": "control_entailed", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "facts_positive": ["Part E has release approval.", "Part E passed the visual check."], "facts_negative": [], "rules_natural": ["If a part passed the visual check and has release approval, then that part may ship to customers."], "predicate_glossary": {"passed_visual_check": "passed the visual check", "has_release_approval": "has release approval", "has_supplier_note": "has a supplier note", "may_ship": "may ship to customers", "needs_engineer_review": "needs engineer review"}, "vocabulary_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "query_atom": "may_ship::Part E", "target_statement": "Part E may ship to customers.", "gold_reason_type": "entailed_by_fact_or_rule", "symbolic": {"positive_atoms": ["has_release_approval::Part E", "passed_visual_check::Part E"], "negative_atoms": [], "rules": [{"antecedents": ["passed_visual_check::Part E", "has_release_approval::Part E"], "conclusion": "may_ship::Part E", "text": "If a part passed the visual check and has release approval, then that part may ship to customers."}], "query_atom": "may_ship::Part E", "closure_atoms": ["has_release_approval::Part E", "may_ship::Part E", "passed_visual_check::Part E"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Part E has release approval.", "Part E passed the visual check."], "negative": []}}, "gold_truth_value": "true", "gold_source_used": "agent_b", "gold_closure_handling": "entailed_or_explicit", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: has_release_approval, has_supplier_note, may_ship, needs_engineer_review, passed_visual_check.\n\nPredicate glossary:\n- has_release_approval: has release approval\n- has_supplier_note: has a supplier note\n- may_ship: may ship to customers\n- needs_engineer_review: needs engineer review\n- passed_visual_check: passed the visual check\n\nCoordinator policy rules:\n- If a part passed the visual check and has release approval, then that part may ship to customers.\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Part E has release approval.\n- Part E passed the visual check.\nNegative facts:\n- none\n\nTarget atom: may_ship::Part E.\nTarget statement: Part E may ship to customers.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "manufacturing_quality_0108_entailed_open_conclusion__cwa__multi_agent", "source_id": "manufacturing_quality_0108_entailed_open_conclusion__cwa", "base_id": "manufacturing_quality_0108_entailed_open_conclusion", "split": "train", "domain": "manufacturing_quality", "family": "entailed_open_conclusion", "subset": "control_entailed", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "agent_a_complete_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "agent_b_open_predicates": [], "facts_positive": ["Part A has a supplier note."], "facts_negative": [], "rules_natural": ["If a part has a supplier note, then that part needs engineer review."], "predicate_glossary": {"passed_visual_check": "passed the visual check", "has_release_approval": "has release approval", "has_supplier_note": "has a supplier note", "may_ship": "may ship to customers", "needs_engineer_review": "needs engineer review"}, "vocabulary_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "query_atom": "needs_engineer_review::Part A", "target_statement": "Part A needs engineer review.", "gold_reason_type": "entailed_by_fact_or_rule", "symbolic": {"positive_atoms": ["has_supplier_note::Part A"], "negative_atoms": [], "rules": [{"antecedents": ["has_supplier_note::Part A"], "conclusion": "needs_engineer_review::Part A", "text": "If a part has a supplier note, then that part needs engineer review."}], "query_atom": "needs_engineer_review::Part A", "closure_atoms": ["has_supplier_note::Part A", "needs_engineer_review::Part A"]}, "source_reports": {"agent_a": {"positive": ["Part A has a supplier note."], "negative": []}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "true", "gold_source_used": "agent_a", "gold_closure_handling": "entailed_or_explicit", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_release_approval, has_supplier_note, may_ship, needs_engineer_review, passed_visual_check.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- has_release_approval: has release approval\n- has_supplier_note: has a supplier note\n- may_ship: may ship to customers\n- needs_engineer_review: needs engineer review\n- passed_visual_check: passed the visual check\n\nCoordinator policy rules:\n- If a part has a supplier note, then that part needs engineer review.\n\nAgent A report:\nPositive facts:\n- Part A has a supplier note.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: needs_engineer_review::Part A.\nTarget statement: Part A needs engineer review.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "manufacturing_quality_0108_entailed_open_conclusion__lcwa__multi_agent", "source_id": "manufacturing_quality_0108_entailed_open_conclusion__lcwa", "base_id": "manufacturing_quality_0108_entailed_open_conclusion", "split": "train", "domain": "manufacturing_quality", "family": "entailed_open_conclusion", "subset": "control_entailed", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_release_approval", "may_ship"], "agent_a_complete_predicates": ["has_release_approval", "may_ship"], "agent_b_open_predicates": ["has_supplier_note", "needs_engineer_review", "passed_visual_check"], "facts_positive": ["Part A has a supplier note."], "facts_negative": [], "rules_natural": ["If a part has a supplier note, then that part needs engineer review."], "predicate_glossary": {"passed_visual_check": "passed the visual check", "has_release_approval": "has release approval", "has_supplier_note": "has a supplier note", "may_ship": "may ship to customers", "needs_engineer_review": "needs engineer review"}, "vocabulary_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "query_atom": "needs_engineer_review::Part A", "target_statement": "Part A needs engineer review.", "gold_reason_type": "entailed_by_fact_or_rule", "symbolic": {"positive_atoms": ["has_supplier_note::Part A"], "negative_atoms": [], "rules": [{"antecedents": ["has_supplier_note::Part A"], "conclusion": "needs_engineer_review::Part A", "text": "If a part has a supplier note, then that part needs engineer review."}], "query_atom": "needs_engineer_review::Part A", "closure_atoms": ["has_supplier_note::Part A", "needs_engineer_review::Part A"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Part A has a supplier note."], "negative": []}}, "gold_truth_value": "true", "gold_source_used": "agent_b", "gold_closure_handling": "entailed_or_explicit", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_release_approval, may_ship.\nAgent B open/incomplete predicates: has_supplier_note, needs_engineer_review, passed_visual_check.\n\nPredicate glossary:\n- has_release_approval: has release approval\n- has_supplier_note: has a supplier note\n- may_ship: may ship to customers\n- needs_engineer_review: needs engineer review\n- passed_visual_check: passed the visual check\n\nCoordinator policy rules:\n- If a part has a supplier note, then that part needs engineer review.\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Part A has a supplier note.\nNegative facts:\n- none\n\nTarget atom: needs_engineer_review::Part A.\nTarget statement: Part A needs engineer review.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "manufacturing_quality_0108_entailed_open_conclusion__owa__multi_agent", "source_id": "manufacturing_quality_0108_entailed_open_conclusion__owa", "base_id": "manufacturing_quality_0108_entailed_open_conclusion", "split": "train", "domain": "manufacturing_quality", "family": "entailed_open_conclusion", "subset": "control_entailed", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "facts_positive": ["Part A has a supplier note."], "facts_negative": [], "rules_natural": ["If a part has a supplier note, then that part needs engineer review."], "predicate_glossary": {"passed_visual_check": "passed the visual check", "has_release_approval": "has release approval", "has_supplier_note": "has a supplier note", "may_ship": "may ship to customers", "needs_engineer_review": "needs engineer review"}, "vocabulary_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "query_atom": "needs_engineer_review::Part A", "target_statement": "Part A needs engineer review.", "gold_reason_type": "entailed_by_fact_or_rule", "symbolic": {"positive_atoms": ["has_supplier_note::Part A"], "negative_atoms": [], "rules": [{"antecedents": ["has_supplier_note::Part A"], "conclusion": "needs_engineer_review::Part A", "text": "If a part has a supplier note, then that part needs engineer review."}], "query_atom": "needs_engineer_review::Part A", "closure_atoms": ["has_supplier_note::Part A", "needs_engineer_review::Part A"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Part A has a supplier note."], "negative": []}}, "gold_truth_value": "true", "gold_source_used": "agent_b", "gold_closure_handling": "entailed_or_explicit", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: has_release_approval, has_supplier_note, may_ship, needs_engineer_review, passed_visual_check.\n\nPredicate glossary:\n- has_release_approval: has release approval\n- has_supplier_note: has a supplier note\n- may_ship: may ship to customers\n- needs_engineer_review: needs engineer review\n- passed_visual_check: passed the visual check\n\nCoordinator policy rules:\n- If a part has a supplier note, then that part needs engineer review.\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Part A has a supplier note.\nNegative facts:\n- none\n\nTarget atom: needs_engineer_review::Part A.\nTarget statement: Part A needs engineer review.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "manufacturing_quality_0109_entailed_open_conclusion__cwa__multi_agent", "source_id": "manufacturing_quality_0109_entailed_open_conclusion__cwa", "base_id": "manufacturing_quality_0109_entailed_open_conclusion", "split": "test", "domain": "manufacturing_quality", "family": "entailed_open_conclusion", "subset": "control_entailed", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "agent_a_complete_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "agent_b_open_predicates": [], "facts_positive": ["Part B has a supplier note."], "facts_negative": [], "rules_natural": ["If a part has a supplier note, then that part needs engineer review."], "predicate_glossary": {"passed_visual_check": "passed the visual check", "has_release_approval": "has release approval", "has_supplier_note": "has a supplier note", "may_ship": "may ship to customers", "needs_engineer_review": "needs engineer review"}, "vocabulary_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "query_atom": "needs_engineer_review::Part B", "target_statement": "Part B needs engineer review.", "gold_reason_type": "entailed_by_fact_or_rule", "symbolic": {"positive_atoms": ["has_supplier_note::Part B"], "negative_atoms": [], "rules": [{"antecedents": ["has_supplier_note::Part B"], "conclusion": "needs_engineer_review::Part B", "text": "If a part has a supplier note, then that part needs engineer review."}], "query_atom": "needs_engineer_review::Part B", "closure_atoms": ["has_supplier_note::Part B", "needs_engineer_review::Part B"]}, "source_reports": {"agent_a": {"positive": ["Part B has a supplier note."], "negative": []}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "true", "gold_source_used": "agent_a", "gold_closure_handling": "entailed_or_explicit", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_release_approval, has_supplier_note, may_ship, needs_engineer_review, passed_visual_check.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- has_release_approval: has release approval\n- has_supplier_note: has a supplier note\n- may_ship: may ship to customers\n- needs_engineer_review: needs engineer review\n- passed_visual_check: passed the visual check\n\nCoordinator policy rules:\n- If a part has a supplier note, then that part needs engineer review.\n\nAgent A report:\nPositive facts:\n- Part B has a supplier note.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: needs_engineer_review::Part B.\nTarget statement: Part B needs engineer review.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "manufacturing_quality_0109_entailed_open_conclusion__lcwa__multi_agent", "source_id": "manufacturing_quality_0109_entailed_open_conclusion__lcwa", "base_id": "manufacturing_quality_0109_entailed_open_conclusion", "split": "test", "domain": "manufacturing_quality", "family": "entailed_open_conclusion", "subset": "control_entailed", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_release_approval", "may_ship"], "agent_a_complete_predicates": ["has_release_approval", "may_ship"], "agent_b_open_predicates": ["has_supplier_note", "needs_engineer_review", "passed_visual_check"], "facts_positive": ["Part B has a supplier note."], "facts_negative": [], "rules_natural": ["If a part has a supplier note, then that part needs engineer review."], "predicate_glossary": {"passed_visual_check": "passed the visual check", "has_release_approval": "has release approval", "has_supplier_note": "has a supplier note", "may_ship": "may ship to customers", "needs_engineer_review": "needs engineer review"}, "vocabulary_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "query_atom": "needs_engineer_review::Part B", "target_statement": "Part B needs engineer review.", "gold_reason_type": "entailed_by_fact_or_rule", "symbolic": {"positive_atoms": ["has_supplier_note::Part B"], "negative_atoms": [], "rules": [{"antecedents": ["has_supplier_note::Part B"], "conclusion": "needs_engineer_review::Part B", "text": "If a part has a supplier note, then that part needs engineer review."}], "query_atom": "needs_engineer_review::Part B", "closure_atoms": ["has_supplier_note::Part B", "needs_engineer_review::Part B"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Part B has a supplier note."], "negative": []}}, "gold_truth_value": "true", "gold_source_used": "agent_b", "gold_closure_handling": "entailed_or_explicit", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_release_approval, may_ship.\nAgent B open/incomplete predicates: has_supplier_note, needs_engineer_review, passed_visual_check.\n\nPredicate glossary:\n- has_release_approval: has release approval\n- has_supplier_note: has a supplier note\n- may_ship: may ship to customers\n- needs_engineer_review: needs engineer review\n- passed_visual_check: passed the visual check\n\nCoordinator policy rules:\n- If a part has a supplier note, then that part needs engineer review.\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Part B has a supplier note.\nNegative facts:\n- none\n\nTarget atom: needs_engineer_review::Part B.\nTarget statement: Part B needs engineer review.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "manufacturing_quality_0109_entailed_open_conclusion__owa__multi_agent", "source_id": "manufacturing_quality_0109_entailed_open_conclusion__owa", "base_id": "manufacturing_quality_0109_entailed_open_conclusion", "split": "test", "domain": "manufacturing_quality", "family": "entailed_open_conclusion", "subset": "control_entailed", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "facts_positive": ["Part B has a supplier note."], "facts_negative": [], "rules_natural": ["If a part has a supplier note, then that part needs engineer review."], "predicate_glossary": {"passed_visual_check": "passed the visual check", "has_release_approval": "has release approval", "has_supplier_note": "has a supplier note", "may_ship": "may ship to customers", "needs_engineer_review": "needs engineer review"}, "vocabulary_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "query_atom": "needs_engineer_review::Part B", "target_statement": "Part B needs engineer review.", "gold_reason_type": "entailed_by_fact_or_rule", "symbolic": {"positive_atoms": ["has_supplier_note::Part B"], "negative_atoms": [], "rules": [{"antecedents": ["has_supplier_note::Part B"], "conclusion": "needs_engineer_review::Part B", "text": "If a part has a supplier note, then that part needs engineer review."}], "query_atom": "needs_engineer_review::Part B", "closure_atoms": ["has_supplier_note::Part B", "needs_engineer_review::Part B"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Part B has a supplier note."], "negative": []}}, "gold_truth_value": "true", "gold_source_used": "agent_b", "gold_closure_handling": "entailed_or_explicit", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: has_release_approval, has_supplier_note, may_ship, needs_engineer_review, passed_visual_check.\n\nPredicate glossary:\n- has_release_approval: has release approval\n- has_supplier_note: has a supplier note\n- may_ship: may ship to customers\n- needs_engineer_review: needs engineer review\n- passed_visual_check: passed the visual check\n\nCoordinator policy rules:\n- If a part has a supplier note, then that part needs engineer review.\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Part B has a supplier note.\nNegative facts:\n- none\n\nTarget atom: needs_engineer_review::Part B.\nTarget statement: Part B needs engineer review.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "manufacturing_quality_0110_entailed_open_conclusion__cwa__multi_agent", "source_id": "manufacturing_quality_0110_entailed_open_conclusion__cwa", "base_id": "manufacturing_quality_0110_entailed_open_conclusion", "split": "train", "domain": "manufacturing_quality", "family": "entailed_open_conclusion", "subset": "control_entailed", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "agent_a_complete_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "agent_b_open_predicates": [], "facts_positive": ["Part C has a supplier note."], "facts_negative": [], "rules_natural": ["If a part has a supplier note, then that part needs engineer review."], "predicate_glossary": {"passed_visual_check": "passed the visual check", "has_release_approval": "has release approval", "has_supplier_note": "has a supplier note", "may_ship": "may ship to customers", "needs_engineer_review": "needs engineer review"}, "vocabulary_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "query_atom": "needs_engineer_review::Part C", "target_statement": "Part C needs engineer review.", "gold_reason_type": "entailed_by_fact_or_rule", "symbolic": {"positive_atoms": ["has_supplier_note::Part C"], "negative_atoms": [], "rules": [{"antecedents": ["has_supplier_note::Part C"], "conclusion": "needs_engineer_review::Part C", "text": "If a part has a supplier note, then that part needs engineer review."}], "query_atom": "needs_engineer_review::Part C", "closure_atoms": ["has_supplier_note::Part C", "needs_engineer_review::Part C"]}, "source_reports": {"agent_a": {"positive": ["Part C has a supplier note."], "negative": []}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "true", "gold_source_used": "agent_a", "gold_closure_handling": "entailed_or_explicit", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_release_approval, has_supplier_note, may_ship, needs_engineer_review, passed_visual_check.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- has_release_approval: has release approval\n- has_supplier_note: has a supplier note\n- may_ship: may ship to customers\n- needs_engineer_review: needs engineer review\n- passed_visual_check: passed the visual check\n\nCoordinator policy rules:\n- If a part has a supplier note, then that part needs engineer review.\n\nAgent A report:\nPositive facts:\n- Part C has a supplier note.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: needs_engineer_review::Part C.\nTarget statement: Part C needs engineer review.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "manufacturing_quality_0110_entailed_open_conclusion__lcwa__multi_agent", "source_id": "manufacturing_quality_0110_entailed_open_conclusion__lcwa", "base_id": "manufacturing_quality_0110_entailed_open_conclusion", "split": "train", "domain": "manufacturing_quality", "family": "entailed_open_conclusion", "subset": "control_entailed", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_release_approval", "may_ship"], "agent_a_complete_predicates": ["has_release_approval", "may_ship"], "agent_b_open_predicates": ["has_supplier_note", "needs_engineer_review", "passed_visual_check"], "facts_positive": ["Part C has a supplier note."], "facts_negative": [], "rules_natural": ["If a part has a supplier note, then that part needs engineer review."], "predicate_glossary": {"passed_visual_check": "passed the visual check", "has_release_approval": "has release approval", "has_supplier_note": "has a supplier note", "may_ship": "may ship to customers", "needs_engineer_review": "needs engineer review"}, "vocabulary_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "query_atom": "needs_engineer_review::Part C", "target_statement": "Part C needs engineer review.", "gold_reason_type": "entailed_by_fact_or_rule", "symbolic": {"positive_atoms": ["has_supplier_note::Part C"], "negative_atoms": [], "rules": [{"antecedents": ["has_supplier_note::Part C"], "conclusion": "needs_engineer_review::Part C", "text": "If a part has a supplier note, then that part needs engineer review."}], "query_atom": "needs_engineer_review::Part C", "closure_atoms": ["has_supplier_note::Part C", "needs_engineer_review::Part C"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Part C has a supplier note."], "negative": []}}, "gold_truth_value": "true", "gold_source_used": "agent_b", "gold_closure_handling": "entailed_or_explicit", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_release_approval, may_ship.\nAgent B open/incomplete predicates: has_supplier_note, needs_engineer_review, passed_visual_check.\n\nPredicate glossary:\n- has_release_approval: has release approval\n- has_supplier_note: has a supplier note\n- may_ship: may ship to customers\n- needs_engineer_review: needs engineer review\n- passed_visual_check: passed the visual check\n\nCoordinator policy rules:\n- If a part has a supplier note, then that part needs engineer review.\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Part C has a supplier note.\nNegative facts:\n- none\n\nTarget atom: needs_engineer_review::Part C.\nTarget statement: Part C needs engineer review.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "manufacturing_quality_0110_entailed_open_conclusion__owa__multi_agent", "source_id": "manufacturing_quality_0110_entailed_open_conclusion__owa", "base_id": "manufacturing_quality_0110_entailed_open_conclusion", "split": "train", "domain": "manufacturing_quality", "family": "entailed_open_conclusion", "subset": "control_entailed", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "facts_positive": ["Part C has a supplier note."], "facts_negative": [], "rules_natural": ["If a part has a supplier note, then that part needs engineer review."], "predicate_glossary": {"passed_visual_check": "passed the visual check", "has_release_approval": "has release approval", "has_supplier_note": "has a supplier note", "may_ship": "may ship to customers", "needs_engineer_review": "needs engineer review"}, "vocabulary_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "query_atom": "needs_engineer_review::Part C", "target_statement": "Part C needs engineer review.", "gold_reason_type": "entailed_by_fact_or_rule", "symbolic": {"positive_atoms": ["has_supplier_note::Part C"], "negative_atoms": [], "rules": [{"antecedents": ["has_supplier_note::Part C"], "conclusion": "needs_engineer_review::Part C", "text": "If a part has a supplier note, then that part needs engineer review."}], "query_atom": "needs_engineer_review::Part C", "closure_atoms": ["has_supplier_note::Part C", "needs_engineer_review::Part C"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Part C has a supplier note."], "negative": []}}, "gold_truth_value": "true", "gold_source_used": "agent_b", "gold_closure_handling": "entailed_or_explicit", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: has_release_approval, has_supplier_note, may_ship, needs_engineer_review, passed_visual_check.\n\nPredicate glossary:\n- has_release_approval: has release approval\n- has_supplier_note: has a supplier note\n- may_ship: may ship to customers\n- needs_engineer_review: needs engineer review\n- passed_visual_check: passed the visual check\n\nCoordinator policy rules:\n- If a part has a supplier note, then that part needs engineer review.\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Part C has a supplier note.\nNegative facts:\n- none\n\nTarget atom: needs_engineer_review::Part C.\nTarget statement: Part C needs engineer review.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "manufacturing_quality_0112_explicit_negative_closed__cwa__multi_agent", "source_id": "manufacturing_quality_0112_explicit_negative_closed__cwa", "base_id": "manufacturing_quality_0112_explicit_negative_closed", "split": "train", "domain": "manufacturing_quality", "family": "explicit_negative_closed", "subset": "control_explicit_negative", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "agent_a_complete_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "agent_b_open_predicates": [], "facts_positive": ["Part A passed the visual check."], "facts_negative": ["Part E does not has release approval."], "rules_natural": [], "predicate_glossary": {"passed_visual_check": "passed the visual check", "has_release_approval": "has release approval", "has_supplier_note": "has a supplier note", "may_ship": "may ship to customers", "needs_engineer_review": "needs engineer review"}, "vocabulary_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "query_atom": "has_release_approval::Part E", "target_statement": "Part E has release approval.", "gold_reason_type": "explicit_negative_fact", "symbolic": {"positive_atoms": ["passed_visual_check::Part A"], "negative_atoms": ["has_release_approval::Part E"], "rules": [], "query_atom": "has_release_approval::Part E", "closure_atoms": ["passed_visual_check::Part A"]}, "source_reports": {"agent_a": {"positive": ["Part A passed the visual check."], "negative": ["Part E does not has release approval."]}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "explicit_negative", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_release_approval, has_supplier_note, may_ship, needs_engineer_review, passed_visual_check.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- has_release_approval: has release approval\n- has_supplier_note: has a supplier note\n- may_ship: may ship to customers\n- needs_engineer_review: needs engineer review\n- passed_visual_check: passed the visual check\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- Part A passed the visual check.\nNegative facts:\n- Part E does not has release approval.\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: has_release_approval::Part E.\nTarget statement: Part E has release approval.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "manufacturing_quality_0112_explicit_negative_closed__lcwa__multi_agent", "source_id": "manufacturing_quality_0112_explicit_negative_closed__lcwa", "base_id": "manufacturing_quality_0112_explicit_negative_closed", "split": "train", "domain": "manufacturing_quality", "family": "explicit_negative_closed", "subset": "control_explicit_negative", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_release_approval"], "agent_a_complete_predicates": ["has_release_approval"], "agent_b_open_predicates": ["has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "facts_positive": ["Part A passed the visual check."], "facts_negative": ["Part E does not has release approval."], "rules_natural": [], "predicate_glossary": {"passed_visual_check": "passed the visual check", "has_release_approval": "has release approval", "has_supplier_note": "has a supplier note", "may_ship": "may ship to customers", "needs_engineer_review": "needs engineer review"}, "vocabulary_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "query_atom": "has_release_approval::Part E", "target_statement": "Part E has release approval.", "gold_reason_type": "explicit_negative_fact", "symbolic": {"positive_atoms": ["passed_visual_check::Part A"], "negative_atoms": ["has_release_approval::Part E"], "rules": [], "query_atom": "has_release_approval::Part E", "closure_atoms": ["passed_visual_check::Part A"]}, "source_reports": {"agent_a": {"positive": [], "negative": ["Part E does not has release approval."]}, "agent_b": {"positive": ["Part A passed the visual check."], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "explicit_negative", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_release_approval.\nAgent B open/incomplete predicates: has_supplier_note, may_ship, needs_engineer_review, passed_visual_check.\n\nPredicate glossary:\n- has_release_approval: has release approval\n- has_supplier_note: has a supplier note\n- may_ship: may ship to customers\n- needs_engineer_review: needs engineer review\n- passed_visual_check: passed the visual check\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- Part E does not has release approval.\n\nAgent B report:\nPositive facts:\n- Part A passed the visual check.\nNegative facts:\n- none\n\nTarget atom: has_release_approval::Part E.\nTarget statement: Part E has release approval.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "manufacturing_quality_0112_explicit_negative_closed__owa__multi_agent", "source_id": "manufacturing_quality_0112_explicit_negative_closed__owa", "base_id": "manufacturing_quality_0112_explicit_negative_closed", "split": "train", "domain": "manufacturing_quality", "family": "explicit_negative_closed", "subset": "control_explicit_negative", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "facts_positive": ["Part A passed the visual check."], "facts_negative": ["Part E does not has release approval."], "rules_natural": [], "predicate_glossary": {"passed_visual_check": "passed the visual check", "has_release_approval": "has release approval", "has_supplier_note": "has a supplier note", "may_ship": "may ship to customers", "needs_engineer_review": "needs engineer review"}, "vocabulary_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "query_atom": "has_release_approval::Part E", "target_statement": "Part E has release approval.", "gold_reason_type": "explicit_negative_fact", "symbolic": {"positive_atoms": ["passed_visual_check::Part A"], "negative_atoms": ["has_release_approval::Part E"], "rules": [], "query_atom": "has_release_approval::Part E", "closure_atoms": ["passed_visual_check::Part A"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Part A passed the visual check."], "negative": ["Part E does not has release approval."]}}, "gold_truth_value": "false", "gold_source_used": "agent_b", "gold_closure_handling": "explicit_negative", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: has_release_approval, has_supplier_note, may_ship, needs_engineer_review, passed_visual_check.\n\nPredicate glossary:\n- has_release_approval: has release approval\n- has_supplier_note: has a supplier note\n- may_ship: may ship to customers\n- needs_engineer_review: needs engineer review\n- passed_visual_check: passed the visual check\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Part A passed the visual check.\nNegative facts:\n- Part E does not has release approval.\n\nTarget atom: has_release_approval::Part E.\nTarget statement: Part E has release approval.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "manufacturing_quality_0114_explicit_negative_closed__cwa__multi_agent", "source_id": "manufacturing_quality_0114_explicit_negative_closed__cwa", "base_id": "manufacturing_quality_0114_explicit_negative_closed", "split": "train", "domain": "manufacturing_quality", "family": "explicit_negative_closed", "subset": "control_explicit_negative", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "agent_a_complete_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "agent_b_open_predicates": [], "facts_positive": ["Part C passed the visual check."], "facts_negative": ["Part A does not has release approval."], "rules_natural": [], "predicate_glossary": {"passed_visual_check": "passed the visual check", "has_release_approval": "has release approval", "has_supplier_note": "has a supplier note", "may_ship": "may ship to customers", "needs_engineer_review": "needs engineer review"}, "vocabulary_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "query_atom": "has_release_approval::Part A", "target_statement": "Part A has release approval.", "gold_reason_type": "explicit_negative_fact", "symbolic": {"positive_atoms": ["passed_visual_check::Part C"], "negative_atoms": ["has_release_approval::Part A"], "rules": [], "query_atom": "has_release_approval::Part A", "closure_atoms": ["passed_visual_check::Part C"]}, "source_reports": {"agent_a": {"positive": ["Part C passed the visual check."], "negative": ["Part A does not has release approval."]}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "explicit_negative", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_release_approval, has_supplier_note, may_ship, needs_engineer_review, passed_visual_check.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- has_release_approval: has release approval\n- has_supplier_note: has a supplier note\n- may_ship: may ship to customers\n- needs_engineer_review: needs engineer review\n- passed_visual_check: passed the visual check\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- Part C passed the visual check.\nNegative facts:\n- Part A does not has release approval.\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: has_release_approval::Part A.\nTarget statement: Part A has release approval.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "manufacturing_quality_0114_explicit_negative_closed__lcwa__multi_agent", "source_id": "manufacturing_quality_0114_explicit_negative_closed__lcwa", "base_id": "manufacturing_quality_0114_explicit_negative_closed", "split": "train", "domain": "manufacturing_quality", "family": "explicit_negative_closed", "subset": "control_explicit_negative", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_release_approval"], "agent_a_complete_predicates": ["has_release_approval"], "agent_b_open_predicates": ["has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "facts_positive": ["Part C passed the visual check."], "facts_negative": ["Part A does not has release approval."], "rules_natural": [], "predicate_glossary": {"passed_visual_check": "passed the visual check", "has_release_approval": "has release approval", "has_supplier_note": "has a supplier note", "may_ship": "may ship to customers", "needs_engineer_review": "needs engineer review"}, "vocabulary_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "query_atom": "has_release_approval::Part A", "target_statement": "Part A has release approval.", "gold_reason_type": "explicit_negative_fact", "symbolic": {"positive_atoms": ["passed_visual_check::Part C"], "negative_atoms": ["has_release_approval::Part A"], "rules": [], "query_atom": "has_release_approval::Part A", "closure_atoms": ["passed_visual_check::Part C"]}, "source_reports": {"agent_a": {"positive": [], "negative": ["Part A does not has release approval."]}, "agent_b": {"positive": ["Part C passed the visual check."], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "explicit_negative", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_release_approval.\nAgent B open/incomplete predicates: has_supplier_note, may_ship, needs_engineer_review, passed_visual_check.\n\nPredicate glossary:\n- has_release_approval: has release approval\n- has_supplier_note: has a supplier note\n- may_ship: may ship to customers\n- needs_engineer_review: needs engineer review\n- passed_visual_check: passed the visual check\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- Part A does not has release approval.\n\nAgent B report:\nPositive facts:\n- Part C passed the visual check.\nNegative facts:\n- none\n\nTarget atom: has_release_approval::Part A.\nTarget statement: Part A has release approval.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "manufacturing_quality_0114_explicit_negative_closed__owa__multi_agent", "source_id": "manufacturing_quality_0114_explicit_negative_closed__owa", "base_id": "manufacturing_quality_0114_explicit_negative_closed", "split": "train", "domain": "manufacturing_quality", "family": "explicit_negative_closed", "subset": "control_explicit_negative", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "facts_positive": ["Part C passed the visual check."], "facts_negative": ["Part A does not has release approval."], "rules_natural": [], "predicate_glossary": {"passed_visual_check": "passed the visual check", "has_release_approval": "has release approval", "has_supplier_note": "has a supplier note", "may_ship": "may ship to customers", "needs_engineer_review": "needs engineer review"}, "vocabulary_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "query_atom": "has_release_approval::Part A", "target_statement": "Part A has release approval.", "gold_reason_type": "explicit_negative_fact", "symbolic": {"positive_atoms": ["passed_visual_check::Part C"], "negative_atoms": ["has_release_approval::Part A"], "rules": [], "query_atom": "has_release_approval::Part A", "closure_atoms": ["passed_visual_check::Part C"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Part C passed the visual check."], "negative": ["Part A does not has release approval."]}}, "gold_truth_value": "false", "gold_source_used": "agent_b", "gold_closure_handling": "explicit_negative", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: has_release_approval, has_supplier_note, may_ship, needs_engineer_review, passed_visual_check.\n\nPredicate glossary:\n- has_release_approval: has release approval\n- has_supplier_note: has a supplier note\n- may_ship: may ship to customers\n- needs_engineer_review: needs engineer review\n- passed_visual_check: passed the visual check\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Part C passed the visual check.\nNegative facts:\n- Part A does not has release approval.\n\nTarget atom: has_release_approval::Part A.\nTarget statement: Part A has release approval.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "manufacturing_quality_0115_explicit_negative_closed__cwa__multi_agent", "source_id": "manufacturing_quality_0115_explicit_negative_closed__cwa", "base_id": "manufacturing_quality_0115_explicit_negative_closed", "split": "test", "domain": "manufacturing_quality", "family": "explicit_negative_closed", "subset": "control_explicit_negative", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "agent_a_complete_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "agent_b_open_predicates": [], "facts_positive": ["Part D passed the visual check."], "facts_negative": ["Part B does not has release approval."], "rules_natural": [], "predicate_glossary": {"passed_visual_check": "passed the visual check", "has_release_approval": "has release approval", "has_supplier_note": "has a supplier note", "may_ship": "may ship to customers", "needs_engineer_review": "needs engineer review"}, "vocabulary_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "query_atom": "has_release_approval::Part B", "target_statement": "Part B has release approval.", "gold_reason_type": "explicit_negative_fact", "symbolic": {"positive_atoms": ["passed_visual_check::Part D"], "negative_atoms": ["has_release_approval::Part B"], "rules": [], "query_atom": "has_release_approval::Part B", "closure_atoms": ["passed_visual_check::Part D"]}, "source_reports": {"agent_a": {"positive": ["Part D passed the visual check."], "negative": ["Part B does not has release approval."]}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "explicit_negative", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_release_approval, has_supplier_note, may_ship, needs_engineer_review, passed_visual_check.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- has_release_approval: has release approval\n- has_supplier_note: has a supplier note\n- may_ship: may ship to customers\n- needs_engineer_review: needs engineer review\n- passed_visual_check: passed the visual check\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- Part D passed the visual check.\nNegative facts:\n- Part B does not has release approval.\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: has_release_approval::Part B.\nTarget statement: Part B has release approval.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "manufacturing_quality_0115_explicit_negative_closed__lcwa__multi_agent", "source_id": "manufacturing_quality_0115_explicit_negative_closed__lcwa", "base_id": "manufacturing_quality_0115_explicit_negative_closed", "split": "test", "domain": "manufacturing_quality", "family": "explicit_negative_closed", "subset": "control_explicit_negative", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_release_approval"], "agent_a_complete_predicates": ["has_release_approval"], "agent_b_open_predicates": ["has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "facts_positive": ["Part D passed the visual check."], "facts_negative": ["Part B does not has release approval."], "rules_natural": [], "predicate_glossary": {"passed_visual_check": "passed the visual check", "has_release_approval": "has release approval", "has_supplier_note": "has a supplier note", "may_ship": "may ship to customers", "needs_engineer_review": "needs engineer review"}, "vocabulary_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "query_atom": "has_release_approval::Part B", "target_statement": "Part B has release approval.", "gold_reason_type": "explicit_negative_fact", "symbolic": {"positive_atoms": ["passed_visual_check::Part D"], "negative_atoms": ["has_release_approval::Part B"], "rules": [], "query_atom": "has_release_approval::Part B", "closure_atoms": ["passed_visual_check::Part D"]}, "source_reports": {"agent_a": {"positive": [], "negative": ["Part B does not has release approval."]}, "agent_b": {"positive": ["Part D passed the visual check."], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "explicit_negative", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_release_approval.\nAgent B open/incomplete predicates: has_supplier_note, may_ship, needs_engineer_review, passed_visual_check.\n\nPredicate glossary:\n- has_release_approval: has release approval\n- has_supplier_note: has a supplier note\n- may_ship: may ship to customers\n- needs_engineer_review: needs engineer review\n- passed_visual_check: passed the visual check\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- Part B does not has release approval.\n\nAgent B report:\nPositive facts:\n- Part D passed the visual check.\nNegative facts:\n- none\n\nTarget atom: has_release_approval::Part B.\nTarget statement: Part B has release approval.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "manufacturing_quality_0115_explicit_negative_closed__owa__multi_agent", "source_id": "manufacturing_quality_0115_explicit_negative_closed__owa", "base_id": "manufacturing_quality_0115_explicit_negative_closed", "split": "test", "domain": "manufacturing_quality", "family": "explicit_negative_closed", "subset": "control_explicit_negative", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "facts_positive": ["Part D passed the visual check."], "facts_negative": ["Part B does not has release approval."], "rules_natural": [], "predicate_glossary": {"passed_visual_check": "passed the visual check", "has_release_approval": "has release approval", "has_supplier_note": "has a supplier note", "may_ship": "may ship to customers", "needs_engineer_review": "needs engineer review"}, "vocabulary_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "query_atom": "has_release_approval::Part B", "target_statement": "Part B has release approval.", "gold_reason_type": "explicit_negative_fact", "symbolic": {"positive_atoms": ["passed_visual_check::Part D"], "negative_atoms": ["has_release_approval::Part B"], "rules": [], "query_atom": "has_release_approval::Part B", "closure_atoms": ["passed_visual_check::Part D"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Part D passed the visual check."], "negative": ["Part B does not has release approval."]}}, "gold_truth_value": "false", "gold_source_used": "agent_b", "gold_closure_handling": "explicit_negative", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: has_release_approval, has_supplier_note, may_ship, needs_engineer_review, passed_visual_check.\n\nPredicate glossary:\n- has_release_approval: has release approval\n- has_supplier_note: has a supplier note\n- may_ship: may ship to customers\n- needs_engineer_review: needs engineer review\n- passed_visual_check: passed the visual check\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Part D passed the visual check.\nNegative facts:\n- Part B does not has release approval.\n\nTarget atom: has_release_approval::Part B.\nTarget statement: Part B has release approval.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "manufacturing_quality_0116_explicit_negative_open__cwa__multi_agent", "source_id": "manufacturing_quality_0116_explicit_negative_open__cwa", "base_id": "manufacturing_quality_0116_explicit_negative_open", "split": "train", "domain": "manufacturing_quality", "family": "explicit_negative_open", "subset": "control_explicit_negative", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "agent_a_complete_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "agent_b_open_predicates": [], "facts_positive": ["Part E passed the visual check."], "facts_negative": ["Part C does not has a supplier note."], "rules_natural": [], "predicate_glossary": {"passed_visual_check": "passed the visual check", "has_release_approval": "has release approval", "has_supplier_note": "has a supplier note", "may_ship": "may ship to customers", "needs_engineer_review": "needs engineer review"}, "vocabulary_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "query_atom": "has_supplier_note::Part C", "target_statement": "Part C has a supplier note.", "gold_reason_type": "explicit_negative_fact", "symbolic": {"positive_atoms": ["passed_visual_check::Part E"], "negative_atoms": ["has_supplier_note::Part C"], "rules": [], "query_atom": "has_supplier_note::Part C", "closure_atoms": ["passed_visual_check::Part E"]}, "source_reports": {"agent_a": {"positive": ["Part E passed the visual check."], "negative": ["Part C does not has a supplier note."]}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "explicit_negative", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_release_approval, has_supplier_note, may_ship, needs_engineer_review, passed_visual_check.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- has_release_approval: has release approval\n- has_supplier_note: has a supplier note\n- may_ship: may ship to customers\n- needs_engineer_review: needs engineer review\n- passed_visual_check: passed the visual check\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- Part E passed the visual check.\nNegative facts:\n- Part C does not has a supplier note.\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: has_supplier_note::Part C.\nTarget statement: Part C has a supplier note.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "manufacturing_quality_0116_explicit_negative_open__lcwa__multi_agent", "source_id": "manufacturing_quality_0116_explicit_negative_open__lcwa", "base_id": "manufacturing_quality_0116_explicit_negative_open", "split": "train", "domain": "manufacturing_quality", "family": "explicit_negative_open", "subset": "control_explicit_negative", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_release_approval", "may_ship"], "agent_a_complete_predicates": ["has_release_approval", "may_ship"], "agent_b_open_predicates": ["has_supplier_note", "needs_engineer_review", "passed_visual_check"], "facts_positive": ["Part E passed the visual check."], "facts_negative": ["Part C does not has a supplier note."], "rules_natural": [], "predicate_glossary": {"passed_visual_check": "passed the visual check", "has_release_approval": "has release approval", "has_supplier_note": "has a supplier note", "may_ship": "may ship to customers", "needs_engineer_review": "needs engineer review"}, "vocabulary_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "query_atom": "has_supplier_note::Part C", "target_statement": "Part C has a supplier note.", "gold_reason_type": "explicit_negative_fact", "symbolic": {"positive_atoms": ["passed_visual_check::Part E"], "negative_atoms": ["has_supplier_note::Part C"], "rules": [], "query_atom": "has_supplier_note::Part C", "closure_atoms": ["passed_visual_check::Part E"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Part E passed the visual check."], "negative": ["Part C does not has a supplier note."]}}, "gold_truth_value": "false", "gold_source_used": "agent_b", "gold_closure_handling": "explicit_negative", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_release_approval, may_ship.\nAgent B open/incomplete predicates: has_supplier_note, needs_engineer_review, passed_visual_check.\n\nPredicate glossary:\n- has_release_approval: has release approval\n- has_supplier_note: has a supplier note\n- may_ship: may ship to customers\n- needs_engineer_review: needs engineer review\n- passed_visual_check: passed the visual check\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Part E passed the visual check.\nNegative facts:\n- Part C does not has a supplier note.\n\nTarget atom: has_supplier_note::Part C.\nTarget statement: Part C has a supplier note.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "manufacturing_quality_0116_explicit_negative_open__owa__multi_agent", "source_id": "manufacturing_quality_0116_explicit_negative_open__owa", "base_id": "manufacturing_quality_0116_explicit_negative_open", "split": "train", "domain": "manufacturing_quality", "family": "explicit_negative_open", "subset": "control_explicit_negative", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "facts_positive": ["Part E passed the visual check."], "facts_negative": ["Part C does not has a supplier note."], "rules_natural": [], "predicate_glossary": {"passed_visual_check": "passed the visual check", "has_release_approval": "has release approval", "has_supplier_note": "has a supplier note", "may_ship": "may ship to customers", "needs_engineer_review": "needs engineer review"}, "vocabulary_predicates": ["has_release_approval", "has_supplier_note", "may_ship", "needs_engineer_review", "passed_visual_check"], "query_atom": "has_supplier_note::Part C", "target_statement": "Part C has a supplier note.", "gold_reason_type": "explicit_negative_fact", "symbolic": {"positive_atoms": ["passed_visual_check::Part E"], "negative_atoms": ["has_supplier_note::Part C"], "rules": [], "query_atom": "has_supplier_note::Part C", "closure_atoms": ["passed_visual_check::Part E"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Part E passed the visual check."], "negative": ["Part C does not has a supplier note."]}}, "gold_truth_value": "false", "gold_source_used": "agent_b", "gold_closure_handling": "explicit_negative", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: has_release_approval, has_supplier_note, may_ship, needs_engineer_review, passed_visual_check.\n\nPredicate glossary:\n- has_release_approval: has release approval\n- has_supplier_note: has a supplier note\n- may_ship: may ship to customers\n- needs_engineer_review: needs engineer review\n- passed_visual_check: passed the visual check\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Part E passed the visual check.\nNegative facts:\n- Part C does not has a supplier note.\n\nTarget atom: has_supplier_note::Part C.\nTarget statement: Part C has a supplier note.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "procurement_review_0280_closed_missing_direct__cwa__multi_agent", "source_id": "procurement_review_0280_closed_missing_direct__cwa", "base_id": "procurement_review_0280_closed_missing_direct", "split": "train", "domain": "procurement_review", "family": "closed_missing_direct", "subset": "core_contrastive", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["has_risk_note", "has_vendor_approval", "may_receive_purchase_order", "needs_procurement_review", "submitted_tax_form"], "agent_a_complete_predicates": ["has_risk_note", "has_vendor_approval", "may_receive_purchase_order", "needs_procurement_review", "submitted_tax_form"], "agent_b_open_predicates": [], "facts_positive": ["Vendor E submitted a tax form."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"submitted_tax_form": "submitted a tax form", "has_vendor_approval": "has vendor approval", "has_risk_note": "has a risk note", "may_receive_purchase_order": "may receive a purchase order", "needs_procurement_review": "needs procurement review"}, "vocabulary_predicates": ["has_risk_note", "has_vendor_approval", "may_receive_purchase_order", "needs_procurement_review", "submitted_tax_form"], "query_atom": "has_vendor_approval::Vendor E", "target_statement": "Vendor E has vendor approval.", "gold_reason_type": "global_closed_world_underivable", "symbolic": {"positive_atoms": ["submitted_tax_form::Vendor E"], "negative_atoms": [], "rules": [], "query_atom": "has_vendor_approval::Vendor E", "closure_atoms": ["submitted_tax_form::Vendor E"]}, "source_reports": {"agent_a": {"positive": ["Vendor E submitted a tax form."], "negative": []}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "closed_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_risk_note, has_vendor_approval, may_receive_purchase_order, needs_procurement_review, submitted_tax_form.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- has_risk_note: has a risk note\n- has_vendor_approval: has vendor approval\n- may_receive_purchase_order: may receive a purchase order\n- needs_procurement_review: needs procurement review\n- submitted_tax_form: submitted a tax form\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- Vendor E submitted a tax form.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: has_vendor_approval::Vendor E.\nTarget statement: Vendor E has vendor approval.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "procurement_review_0280_closed_missing_direct__lcwa__multi_agent", "source_id": "procurement_review_0280_closed_missing_direct__lcwa", "base_id": "procurement_review_0280_closed_missing_direct", "split": "train", "domain": "procurement_review", "family": "closed_missing_direct", "subset": "core_contrastive", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_vendor_approval"], "agent_a_complete_predicates": ["has_vendor_approval"], "agent_b_open_predicates": ["has_risk_note", "may_receive_purchase_order", "needs_procurement_review", "submitted_tax_form"], "facts_positive": ["Vendor E submitted a tax form."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"submitted_tax_form": "submitted a tax form", "has_vendor_approval": "has vendor approval", "has_risk_note": "has a risk note", "may_receive_purchase_order": "may receive a purchase order", "needs_procurement_review": "needs procurement review"}, "vocabulary_predicates": ["has_risk_note", "has_vendor_approval", "may_receive_purchase_order", "needs_procurement_review", "submitted_tax_form"], "query_atom": "has_vendor_approval::Vendor E", "target_statement": "Vendor E has vendor approval.", "gold_reason_type": "local_closed_world_underivable", "symbolic": {"positive_atoms": ["submitted_tax_form::Vendor E"], "negative_atoms": [], "rules": [], "query_atom": "has_vendor_approval::Vendor E", "closure_atoms": ["submitted_tax_form::Vendor E"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Vendor E submitted a tax form."], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "closed_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_vendor_approval.\nAgent B open/incomplete predicates: has_risk_note, may_receive_purchase_order, needs_procurement_review, submitted_tax_form.\n\nPredicate glossary:\n- has_risk_note: has a risk note\n- has_vendor_approval: has vendor approval\n- may_receive_purchase_order: may receive a purchase order\n- needs_procurement_review: needs procurement review\n- submitted_tax_form: submitted a tax form\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Vendor E submitted a tax form.\nNegative facts:\n- none\n\nTarget atom: has_vendor_approval::Vendor E.\nTarget statement: Vendor E has vendor approval.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "procurement_review_0280_closed_missing_direct__owa__multi_agent", "source_id": "procurement_review_0280_closed_missing_direct__owa", "base_id": "procurement_review_0280_closed_missing_direct", "split": "train", "domain": "procurement_review", "family": "closed_missing_direct", "subset": "core_contrastive", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["has_risk_note", "has_vendor_approval", "may_receive_purchase_order", "needs_procurement_review", "submitted_tax_form"], "facts_positive": ["Vendor E submitted a tax form."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"submitted_tax_form": "submitted a tax form", "has_vendor_approval": "has vendor approval", "has_risk_note": "has a risk note", "may_receive_purchase_order": "may receive a purchase order", "needs_procurement_review": "needs procurement review"}, "vocabulary_predicates": ["has_risk_note", "has_vendor_approval", "may_receive_purchase_order", "needs_procurement_review", "submitted_tax_form"], "query_atom": "has_vendor_approval::Vendor E", "target_statement": "Vendor E has vendor approval.", "gold_reason_type": "open_world_underivable", "symbolic": {"positive_atoms": ["submitted_tax_form::Vendor E"], "negative_atoms": [], "rules": [], "query_atom": "has_vendor_approval::Vendor E", "closure_atoms": ["submitted_tax_form::Vendor E"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Vendor E submitted a tax form."], "negative": []}}, "gold_truth_value": "unknown", "gold_source_used": "agent_b", "gold_closure_handling": "open_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: has_risk_note, has_vendor_approval, may_receive_purchase_order, needs_procurement_review, submitted_tax_form.\n\nPredicate glossary:\n- has_risk_note: has a risk note\n- has_vendor_approval: has vendor approval\n- may_receive_purchase_order: may receive a purchase order\n- needs_procurement_review: needs procurement review\n- submitted_tax_form: submitted a tax form\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Vendor E submitted a tax form.\nNegative facts:\n- none\n\nTarget atom: has_vendor_approval::Vendor E.\nTarget statement: Vendor E has vendor approval.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "procurement_review_0292_open_missing_with_closed_distractor__cwa__multi_agent", "source_id": "procurement_review_0292_open_missing_with_closed_distractor__cwa", "base_id": "procurement_review_0292_open_missing_with_closed_distractor", "split": "train", "domain": "procurement_review", "family": "open_missing_with_closed_distractor", "subset": "core_contrastive", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["has_risk_note", "has_vendor_approval", "may_receive_purchase_order", "needs_procurement_review", "submitted_tax_form"], "agent_a_complete_predicates": ["has_risk_note", "has_vendor_approval", "may_receive_purchase_order", "needs_procurement_review", "submitted_tax_form"], "agent_b_open_predicates": [], "facts_positive": ["Vendor F has vendor approval.", "Vendor F submitted a tax form."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"submitted_tax_form": "submitted a tax form", "has_vendor_approval": "has vendor approval", "has_risk_note": "has a risk note", "may_receive_purchase_order": "may receive a purchase order", "needs_procurement_review": "needs procurement review"}, "vocabulary_predicates": ["has_risk_note", "has_vendor_approval", "may_receive_purchase_order", "needs_procurement_review", "submitted_tax_form"], "query_atom": "has_risk_note::Vendor E", "target_statement": "Vendor E has a risk note.", "gold_reason_type": "global_closed_world_underivable", "symbolic": {"positive_atoms": ["has_vendor_approval::Vendor F", "submitted_tax_form::Vendor F"], "negative_atoms": [], "rules": [], "query_atom": "has_risk_note::Vendor E", "closure_atoms": ["has_vendor_approval::Vendor F", "submitted_tax_form::Vendor F"]}, "source_reports": {"agent_a": {"positive": ["Vendor F has vendor approval.", "Vendor F submitted a tax form."], "negative": []}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "closed_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_risk_note, has_vendor_approval, may_receive_purchase_order, needs_procurement_review, submitted_tax_form.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- has_risk_note: has a risk note\n- has_vendor_approval: has vendor approval\n- may_receive_purchase_order: may receive a purchase order\n- needs_procurement_review: needs procurement review\n- submitted_tax_form: submitted a tax form\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- Vendor F has vendor approval.\n- Vendor F submitted a tax form.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: has_risk_note::Vendor E.\nTarget statement: Vendor E has a risk note.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "procurement_review_0292_open_missing_with_closed_distractor__lcwa__multi_agent", "source_id": "procurement_review_0292_open_missing_with_closed_distractor__lcwa", "base_id": "procurement_review_0292_open_missing_with_closed_distractor", "split": "train", "domain": "procurement_review", "family": "open_missing_with_closed_distractor", "subset": "core_contrastive", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_vendor_approval", "may_receive_purchase_order"], "agent_a_complete_predicates": ["has_vendor_approval", "may_receive_purchase_order"], "agent_b_open_predicates": ["has_risk_note", "needs_procurement_review", "submitted_tax_form"], "facts_positive": ["Vendor F has vendor approval.", "Vendor F submitted a tax form."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"submitted_tax_form": "submitted a tax form", "has_vendor_approval": "has vendor approval", "has_risk_note": "has a risk note", "may_receive_purchase_order": "may receive a purchase order", "needs_procurement_review": "needs procurement review"}, "vocabulary_predicates": ["has_risk_note", "has_vendor_approval", "may_receive_purchase_order", "needs_procurement_review", "submitted_tax_form"], "query_atom": "has_risk_note::Vendor E", "target_statement": "Vendor E has a risk note.", "gold_reason_type": "local_open_world_underivable", "symbolic": {"positive_atoms": ["has_vendor_approval::Vendor F", "submitted_tax_form::Vendor F"], "negative_atoms": [], "rules": [], "query_atom": "has_risk_note::Vendor E", "closure_atoms": ["has_vendor_approval::Vendor F", "submitted_tax_form::Vendor F"]}, "source_reports": {"agent_a": {"positive": ["Vendor F has vendor approval."], "negative": []}, "agent_b": {"positive": ["Vendor F submitted a tax form."], "negative": []}}, "gold_truth_value": "unknown", "gold_source_used": "agent_b", "gold_closure_handling": "open_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_vendor_approval, may_receive_purchase_order.\nAgent B open/incomplete predicates: has_risk_note, needs_procurement_review, submitted_tax_form.\n\nPredicate glossary:\n- has_risk_note: has a risk note\n- has_vendor_approval: has vendor approval\n- may_receive_purchase_order: may receive a purchase order\n- needs_procurement_review: needs procurement review\n- submitted_tax_form: submitted a tax form\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- Vendor F has vendor approval.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Vendor F submitted a tax form.\nNegative facts:\n- none\n\nTarget atom: has_risk_note::Vendor E.\nTarget statement: Vendor E has a risk note.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "procurement_review_0292_open_missing_with_closed_distractor__owa__multi_agent", "source_id": "procurement_review_0292_open_missing_with_closed_distractor__owa", "base_id": "procurement_review_0292_open_missing_with_closed_distractor", "split": "train", "domain": "procurement_review", "family": "open_missing_with_closed_distractor", "subset": "core_contrastive", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["has_risk_note", "has_vendor_approval", "may_receive_purchase_order", "needs_procurement_review", "submitted_tax_form"], "facts_positive": ["Vendor F has vendor approval.", "Vendor F submitted a tax form."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"submitted_tax_form": "submitted a tax form", "has_vendor_approval": "has vendor approval", "has_risk_note": "has a risk note", "may_receive_purchase_order": "may receive a purchase order", "needs_procurement_review": "needs procurement review"}, "vocabulary_predicates": ["has_risk_note", "has_vendor_approval", "may_receive_purchase_order", "needs_procurement_review", "submitted_tax_form"], "query_atom": "has_risk_note::Vendor E", "target_statement": "Vendor E has a risk note.", "gold_reason_type": "open_world_underivable", "symbolic": {"positive_atoms": ["has_vendor_approval::Vendor F", "submitted_tax_form::Vendor F"], "negative_atoms": [], "rules": [], "query_atom": "has_risk_note::Vendor E", "closure_atoms": ["has_vendor_approval::Vendor F", "submitted_tax_form::Vendor F"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Vendor F has vendor approval.", "Vendor F submitted a tax form."], "negative": []}}, "gold_truth_value": "unknown", "gold_source_used": "agent_b", "gold_closure_handling": "open_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: has_risk_note, has_vendor_approval, may_receive_purchase_order, needs_procurement_review, submitted_tax_form.\n\nPredicate glossary:\n- has_risk_note: has a risk note\n- has_vendor_approval: has vendor approval\n- may_receive_purchase_order: may receive a purchase order\n- needs_procurement_review: needs procurement review\n- submitted_tax_form: submitted a tax form\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Vendor F has vendor approval.\n- Vendor F submitted a tax form.\nNegative facts:\n- none\n\nTarget atom: has_risk_note::Vendor E.\nTarget statement: Vendor E has a risk note.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "procurement_review_0293_open_missing_with_closed_distractor__cwa__multi_agent", "source_id": "procurement_review_0293_open_missing_with_closed_distractor__cwa", "base_id": "procurement_review_0293_open_missing_with_closed_distractor", "split": "train", "domain": "procurement_review", "family": "open_missing_with_closed_distractor", "subset": "core_contrastive", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["has_risk_note", "has_vendor_approval", "may_receive_purchase_order", "needs_procurement_review", "submitted_tax_form"], "agent_a_complete_predicates": ["has_risk_note", "has_vendor_approval", "may_receive_purchase_order", "needs_procurement_review", "submitted_tax_form"], "agent_b_open_predicates": [], "facts_positive": ["Vendor A has vendor approval.", "Vendor A submitted a tax form."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"submitted_tax_form": "submitted a tax form", "has_vendor_approval": "has vendor approval", "has_risk_note": "has a risk note", "may_receive_purchase_order": "may receive a purchase order", "needs_procurement_review": "needs procurement review"}, "vocabulary_predicates": ["has_risk_note", "has_vendor_approval", "may_receive_purchase_order", "needs_procurement_review", "submitted_tax_form"], "query_atom": "has_risk_note::Vendor F", "target_statement": "Vendor F has a risk note.", "gold_reason_type": "global_closed_world_underivable", "symbolic": {"positive_atoms": ["has_vendor_approval::Vendor A", "submitted_tax_form::Vendor A"], "negative_atoms": [], "rules": [], "query_atom": "has_risk_note::Vendor F", "closure_atoms": ["has_vendor_approval::Vendor A", "submitted_tax_form::Vendor A"]}, "source_reports": {"agent_a": {"positive": ["Vendor A has vendor approval.", "Vendor A submitted a tax form."], "negative": []}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "closed_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_risk_note, has_vendor_approval, may_receive_purchase_order, needs_procurement_review, submitted_tax_form.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- has_risk_note: has a risk note\n- has_vendor_approval: has vendor approval\n- may_receive_purchase_order: may receive a purchase order\n- needs_procurement_review: needs procurement review\n- submitted_tax_form: submitted a tax form\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- Vendor A has vendor approval.\n- Vendor A submitted a tax form.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: has_risk_note::Vendor F.\nTarget statement: Vendor F has a risk note.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "procurement_review_0293_open_missing_with_closed_distractor__lcwa__multi_agent", "source_id": "procurement_review_0293_open_missing_with_closed_distractor__lcwa", "base_id": "procurement_review_0293_open_missing_with_closed_distractor", "split": "train", "domain": "procurement_review", "family": "open_missing_with_closed_distractor", "subset": "core_contrastive", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_vendor_approval", "may_receive_purchase_order"], "agent_a_complete_predicates": ["has_vendor_approval", "may_receive_purchase_order"], "agent_b_open_predicates": ["has_risk_note", "needs_procurement_review", "submitted_tax_form"], "facts_positive": ["Vendor A has vendor approval.", "Vendor A submitted a tax form."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"submitted_tax_form": "submitted a tax form", "has_vendor_approval": "has vendor approval", "has_risk_note": "has a risk note", "may_receive_purchase_order": "may receive a purchase order", "needs_procurement_review": "needs procurement review"}, "vocabulary_predicates": ["has_risk_note", "has_vendor_approval", "may_receive_purchase_order", "needs_procurement_review", "submitted_tax_form"], "query_atom": "has_risk_note::Vendor F", "target_statement": "Vendor F has a risk note.", "gold_reason_type": "local_open_world_underivable", "symbolic": {"positive_atoms": ["has_vendor_approval::Vendor A", "submitted_tax_form::Vendor A"], "negative_atoms": [], "rules": [], "query_atom": "has_risk_note::Vendor F", "closure_atoms": ["has_vendor_approval::Vendor A", "submitted_tax_form::Vendor A"]}, "source_reports": {"agent_a": {"positive": ["Vendor A has vendor approval."], "negative": []}, "agent_b": {"positive": ["Vendor A submitted a tax form."], "negative": []}}, "gold_truth_value": "unknown", "gold_source_used": "agent_b", "gold_closure_handling": "open_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_vendor_approval, may_receive_purchase_order.\nAgent B open/incomplete predicates: has_risk_note, needs_procurement_review, submitted_tax_form.\n\nPredicate glossary:\n- has_risk_note: has a risk note\n- has_vendor_approval: has vendor approval\n- may_receive_purchase_order: may receive a purchase order\n- needs_procurement_review: needs procurement review\n- submitted_tax_form: submitted a tax form\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- Vendor A has vendor approval.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Vendor A submitted a tax form.\nNegative facts:\n- none\n\nTarget atom: has_risk_note::Vendor F.\nTarget statement: Vendor F has a risk note.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "procurement_review_0293_open_missing_with_closed_distractor__owa__multi_agent", "source_id": "procurement_review_0293_open_missing_with_closed_distractor__owa", "base_id": "procurement_review_0293_open_missing_with_closed_distractor", "split": "train", "domain": "procurement_review", "family": "open_missing_with_closed_distractor", "subset": "core_contrastive", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["has_risk_note", "has_vendor_approval", "may_receive_purchase_order", "needs_procurement_review", "submitted_tax_form"], "facts_positive": ["Vendor A has vendor approval.", "Vendor A submitted a tax form."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"submitted_tax_form": "submitted a tax form", "has_vendor_approval": "has vendor approval", "has_risk_note": "has a risk note", "may_receive_purchase_order": "may receive a purchase order", "needs_procurement_review": "needs procurement review"}, "vocabulary_predicates": ["has_risk_note", "has_vendor_approval", "may_receive_purchase_order", "needs_procurement_review", "submitted_tax_form"], "query_atom": "has_risk_note::Vendor F", "target_statement": "Vendor F has a risk note.", "gold_reason_type": "open_world_underivable", "symbolic": {"positive_atoms": ["has_vendor_approval::Vendor A", "submitted_tax_form::Vendor A"], "negative_atoms": [], "rules": [], "query_atom": "has_risk_note::Vendor F", "closure_atoms": ["has_vendor_approval::Vendor A", "submitted_tax_form::Vendor A"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Vendor A has vendor approval.", "Vendor A submitted a tax form."], "negative": []}}, "gold_truth_value": "unknown", "gold_source_used": "agent_b", "gold_closure_handling": "open_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: has_risk_note, has_vendor_approval, may_receive_purchase_order, needs_procurement_review, submitted_tax_form.\n\nPredicate glossary:\n- has_risk_note: has a risk note\n- has_vendor_approval: has vendor approval\n- may_receive_purchase_order: may receive a purchase order\n- needs_procurement_review: needs procurement review\n- submitted_tax_form: submitted a tax form\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Vendor A has vendor approval.\n- Vendor A submitted a tax form.\nNegative facts:\n- none\n\nTarget atom: has_risk_note::Vendor F.\nTarget statement: Vendor F has a risk note.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "procurement_review_0294_open_missing_with_closed_distractor__cwa__multi_agent", "source_id": "procurement_review_0294_open_missing_with_closed_distractor__cwa", "base_id": "procurement_review_0294_open_missing_with_closed_distractor", "split": "train", "domain": "procurement_review", "family": "open_missing_with_closed_distractor", "subset": "core_contrastive", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["has_risk_note", "has_vendor_approval", "may_receive_purchase_order", "needs_procurement_review", "submitted_tax_form"], "agent_a_complete_predicates": ["has_risk_note", "has_vendor_approval", "may_receive_purchase_order", "needs_procurement_review", "submitted_tax_form"], "agent_b_open_predicates": [], "facts_positive": ["Vendor B has vendor approval.", "Vendor B submitted a tax form."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"submitted_tax_form": "submitted a tax form", "has_vendor_approval": "has vendor approval", "has_risk_note": "has a risk note", "may_receive_purchase_order": "may receive a purchase order", "needs_procurement_review": "needs procurement review"}, "vocabulary_predicates": ["has_risk_note", "has_vendor_approval", "may_receive_purchase_order", "needs_procurement_review", "submitted_tax_form"], "query_atom": "has_risk_note::Vendor A", "target_statement": "Vendor A has a risk note.", "gold_reason_type": "global_closed_world_underivable", "symbolic": {"positive_atoms": ["has_vendor_approval::Vendor B", "submitted_tax_form::Vendor B"], "negative_atoms": [], "rules": [], "query_atom": "has_risk_note::Vendor A", "closure_atoms": ["has_vendor_approval::Vendor B", "submitted_tax_form::Vendor B"]}, "source_reports": {"agent_a": {"positive": ["Vendor B has vendor approval.", "Vendor B submitted a tax form."], "negative": []}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "closed_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_risk_note, has_vendor_approval, may_receive_purchase_order, needs_procurement_review, submitted_tax_form.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- has_risk_note: has a risk note\n- has_vendor_approval: has vendor approval\n- may_receive_purchase_order: may receive a purchase order\n- needs_procurement_review: needs procurement review\n- submitted_tax_form: submitted a tax form\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- Vendor B has vendor approval.\n- Vendor B submitted a tax form.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: has_risk_note::Vendor A.\nTarget statement: Vendor A has a risk note.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "procurement_review_0294_open_missing_with_closed_distractor__lcwa__multi_agent", "source_id": "procurement_review_0294_open_missing_with_closed_distractor__lcwa", "base_id": "procurement_review_0294_open_missing_with_closed_distractor", "split": "train", "domain": "procurement_review", "family": "open_missing_with_closed_distractor", "subset": "core_contrastive", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_vendor_approval", "may_receive_purchase_order"], "agent_a_complete_predicates": ["has_vendor_approval", "may_receive_purchase_order"], "agent_b_open_predicates": ["has_risk_note", "needs_procurement_review", "submitted_tax_form"], "facts_positive": ["Vendor B has vendor approval.", "Vendor B submitted a tax form."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"submitted_tax_form": "submitted a tax form", "has_vendor_approval": "has vendor approval", "has_risk_note": "has a risk note", "may_receive_purchase_order": "may receive a purchase order", "needs_procurement_review": "needs procurement review"}, "vocabulary_predicates": ["has_risk_note", "has_vendor_approval", "may_receive_purchase_order", "needs_procurement_review", "submitted_tax_form"], "query_atom": "has_risk_note::Vendor A", "target_statement": "Vendor A has a risk note.", "gold_reason_type": "local_open_world_underivable", "symbolic": {"positive_atoms": ["has_vendor_approval::Vendor B", "submitted_tax_form::Vendor B"], "negative_atoms": [], "rules": [], "query_atom": "has_risk_note::Vendor A", "closure_atoms": ["has_vendor_approval::Vendor B", "submitted_tax_form::Vendor B"]}, "source_reports": {"agent_a": {"positive": ["Vendor B has vendor approval."], "negative": []}, "agent_b": {"positive": ["Vendor B submitted a tax form."], "negative": []}}, "gold_truth_value": "unknown", "gold_source_used": "agent_b", "gold_closure_handling": "open_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_vendor_approval, may_receive_purchase_order.\nAgent B open/incomplete predicates: has_risk_note, needs_procurement_review, submitted_tax_form.\n\nPredicate glossary:\n- has_risk_note: has a risk note\n- has_vendor_approval: has vendor approval\n- may_receive_purchase_order: may receive a purchase order\n- needs_procurement_review: needs procurement review\n- submitted_tax_form: submitted a tax form\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- Vendor B has vendor approval.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Vendor B submitted a tax form.\nNegative facts:\n- none\n\nTarget atom: has_risk_note::Vendor A.\nTarget statement: Vendor A has a risk note.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "procurement_review_0294_open_missing_with_closed_distractor__owa__multi_agent", "source_id": "procurement_review_0294_open_missing_with_closed_distractor__owa", "base_id": "procurement_review_0294_open_missing_with_closed_distractor", "split": "train", "domain": "procurement_review", "family": "open_missing_with_closed_distractor", "subset": "core_contrastive", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["has_risk_note", "has_vendor_approval", "may_receive_purchase_order", "needs_procurement_review", "submitted_tax_form"], "facts_positive": ["Vendor B has vendor approval.", "Vendor B submitted a tax form."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"submitted_tax_form": "submitted a tax form", "has_vendor_approval": "has vendor approval", "has_risk_note": "has a risk note", "may_receive_purchase_order": "may receive a purchase order", "needs_procurement_review": "needs procurement review"}, "vocabulary_predicates": ["has_risk_note", "has_vendor_approval", "may_receive_purchase_order", "needs_procurement_review", "submitted_tax_form"], "query_atom": "has_risk_note::Vendor A", "target_statement": "Vendor A has a risk note.", "gold_reason_type": "open_world_underivable", "symbolic": {"positive_atoms": ["has_vendor_approval::Vendor B", "submitted_tax_form::Vendor B"], "negative_atoms": [], "rules": [], "query_atom": "has_risk_note::Vendor A", "closure_atoms": ["has_vendor_approval::Vendor B", "submitted_tax_form::Vendor B"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Vendor B has vendor approval.", "Vendor B submitted a tax form."], "negative": []}}, "gold_truth_value": "unknown", "gold_source_used": "agent_b", "gold_closure_handling": "open_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: has_risk_note, has_vendor_approval, may_receive_purchase_order, needs_procurement_review, submitted_tax_form.\n\nPredicate glossary:\n- has_risk_note: has a risk note\n- has_vendor_approval: has vendor approval\n- may_receive_purchase_order: may receive a purchase order\n- needs_procurement_review: needs procurement review\n- submitted_tax_form: submitted a tax form\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Vendor B has vendor approval.\n- Vendor B submitted a tax form.\nNegative facts:\n- none\n\nTarget atom: has_risk_note::Vendor A.\nTarget statement: Vendor A has a risk note.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "procurement_review_0296_closed_derived_missing_antecedent__cwa__multi_agent", "source_id": "procurement_review_0296_closed_derived_missing_antecedent__cwa", "base_id": "procurement_review_0296_closed_derived_missing_antecedent", "split": "test", "domain": "procurement_review", "family": "closed_derived_missing_antecedent", "subset": "core_contrastive", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["has_risk_note", "has_vendor_approval", "may_receive_purchase_order", "needs_procurement_review", "submitted_tax_form"], "agent_a_complete_predicates": ["has_risk_note", "has_vendor_approval", "may_receive_purchase_order", "needs_procurement_review", "submitted_tax_form"], "agent_b_open_predicates": [], "facts_positive": ["Vendor C submitted a tax form."], "facts_negative": [], "rules_natural": ["If a vendor submitted a tax form and has vendor approval, then that vendor may receive a purchase order."], "predicate_glossary": {"submitted_tax_form": "submitted a tax form", "has_vendor_approval": "has vendor approval", "has_risk_note": "has a risk note", "may_receive_purchase_order": "may receive a purchase order", "needs_procurement_review": "needs procurement review"}, "vocabulary_predicates": ["has_risk_note", "has_vendor_approval", "may_receive_purchase_order", "needs_procurement_review", "submitted_tax_form"], "query_atom": "may_receive_purchase_order::Vendor C", "target_statement": "Vendor C may receive a purchase order.", "gold_reason_type": "global_closed_world_underivable", "symbolic": {"positive_atoms": ["submitted_tax_form::Vendor C"], "negative_atoms": [], "rules": [{"antecedents": ["submitted_tax_form::Vendor C", "has_vendor_approval::Vendor C"], "conclusion": "may_receive_purchase_order::Vendor C", "text": "If a vendor submitted a tax form and has vendor approval, then that vendor may receive a purchase order."}], "query_atom": "may_receive_purchase_order::Vendor C", "closure_atoms": ["submitted_tax_form::Vendor C"]}, "source_reports": {"agent_a": {"positive": ["Vendor C submitted a tax form."], "negative": []}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "closed_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_risk_note, has_vendor_approval, may_receive_purchase_order, needs_procurement_review, submitted_tax_form.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- has_risk_note: has a risk note\n- has_vendor_approval: has vendor approval\n- may_receive_purchase_order: may receive a purchase order\n- needs_procurement_review: needs procurement review\n- submitted_tax_form: submitted a tax form\n\nCoordinator policy rules:\n- If a vendor submitted a tax form and has vendor approval, then that vendor may receive a purchase order.\n\nAgent A report:\nPositive facts:\n- Vendor C submitted a tax form.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: may_receive_purchase_order::Vendor C.\nTarget statement: Vendor C may receive a purchase order.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "procurement_review_0296_closed_derived_missing_antecedent__lcwa__multi_agent", "source_id": "procurement_review_0296_closed_derived_missing_antecedent__lcwa", "base_id": "procurement_review_0296_closed_derived_missing_antecedent", "split": "test", "domain": "procurement_review", "family": "closed_derived_missing_antecedent", "subset": "core_contrastive", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_vendor_approval", "may_receive_purchase_order"], "agent_a_complete_predicates": ["has_vendor_approval", "may_receive_purchase_order"], "agent_b_open_predicates": ["has_risk_note", "needs_procurement_review", "submitted_tax_form"], "facts_positive": ["Vendor C submitted a tax form."], "facts_negative": [], "rules_natural": ["If a vendor submitted a tax form and has vendor approval, then that vendor may receive a purchase order."], "predicate_glossary": {"submitted_tax_form": "submitted a tax form", "has_vendor_approval": "has vendor approval", "has_risk_note": "has a risk note", "may_receive_purchase_order": "may receive a purchase order", "needs_procurement_review": "needs procurement review"}, "vocabulary_predicates": ["has_risk_note", "has_vendor_approval", "may_receive_purchase_order", "needs_procurement_review", "submitted_tax_form"], "query_atom": "may_receive_purchase_order::Vendor C", "target_statement": "Vendor C may receive a purchase order.", "gold_reason_type": "local_closed_world_underivable", "symbolic": {"positive_atoms": ["submitted_tax_form::Vendor C"], "negative_atoms": [], "rules": [{"antecedents": ["submitted_tax_form::Vendor C", "has_vendor_approval::Vendor C"], "conclusion": "may_receive_purchase_order::Vendor C", "text": "If a vendor submitted a tax form and has vendor approval, then that vendor may receive a purchase order."}], "query_atom": "may_receive_purchase_order::Vendor C", "closure_atoms": ["submitted_tax_form::Vendor C"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Vendor C submitted a tax form."], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "closed_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_vendor_approval, may_receive_purchase_order.\nAgent B open/incomplete predicates: has_risk_note, needs_procurement_review, submitted_tax_form.\n\nPredicate glossary:\n- has_risk_note: has a risk note\n- has_vendor_approval: has vendor approval\n- may_receive_purchase_order: may receive a purchase order\n- needs_procurement_review: needs procurement review\n- submitted_tax_form: submitted a tax form\n\nCoordinator policy rules:\n- If a vendor submitted a tax form and has vendor approval, then that vendor may receive a purchase order.\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Vendor C submitted a tax form.\nNegative facts:\n- none\n\nTarget atom: may_receive_purchase_order::Vendor C.\nTarget statement: Vendor C may receive a purchase order.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "procurement_review_0296_closed_derived_missing_antecedent__owa__multi_agent", "source_id": "procurement_review_0296_closed_derived_missing_antecedent__owa", "base_id": "procurement_review_0296_closed_derived_missing_antecedent", "split": "test", "domain": "procurement_review", "family": "closed_derived_missing_antecedent", "subset": "core_contrastive", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["has_risk_note", "has_vendor_approval", "may_receive_purchase_order", "needs_procurement_review", "submitted_tax_form"], "facts_positive": ["Vendor C submitted a tax form."], "facts_negative": [], "rules_natural": ["If a vendor submitted a tax form and has vendor approval, then that vendor may receive a purchase order."], "predicate_glossary": {"submitted_tax_form": "submitted a tax form", "has_vendor_approval": "has vendor approval", "has_risk_note": "has a risk note", "may_receive_purchase_order": "may receive a purchase order", "needs_procurement_review": "needs procurement review"}, "vocabulary_predicates": ["has_risk_note", "has_vendor_approval", "may_receive_purchase_order", "needs_procurement_review", "submitted_tax_form"], "query_atom": "may_receive_purchase_order::Vendor C", "target_statement": "Vendor C may receive a purchase order.", "gold_reason_type": "open_world_underivable", "symbolic": {"positive_atoms": ["submitted_tax_form::Vendor C"], "negative_atoms": [], "rules": [{"antecedents": ["submitted_tax_form::Vendor C", "has_vendor_approval::Vendor C"], "conclusion": "may_receive_purchase_order::Vendor C", "text": "If a vendor submitted a tax form and has vendor approval, then that vendor may receive a purchase order."}], "query_atom": "may_receive_purchase_order::Vendor C", "closure_atoms": ["submitted_tax_form::Vendor C"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Vendor C submitted a tax form."], "negative": []}}, "gold_truth_value": "unknown", "gold_source_used": "agent_b", "gold_closure_handling": "open_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: has_risk_note, has_vendor_approval, may_receive_purchase_order, needs_procurement_review, submitted_tax_form.\n\nPredicate glossary:\n- has_risk_note: has a risk note\n- has_vendor_approval: has vendor approval\n- may_receive_purchase_order: may receive a purchase order\n- needs_procurement_review: needs procurement review\n- submitted_tax_form: submitted a tax form\n\nCoordinator policy rules:\n- If a vendor submitted a tax form and has vendor approval, then that vendor may receive a purchase order.\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Vendor C submitted a tax form.\nNegative facts:\n- none\n\nTarget atom: may_receive_purchase_order::Vendor C.\nTarget statement: Vendor C may receive a purchase order.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "procurement_review_0300_open_derived_missing_antecedent__cwa__multi_agent", "source_id": "procurement_review_0300_open_derived_missing_antecedent__cwa", "base_id": "procurement_review_0300_open_derived_missing_antecedent", "split": "train", "domain": "procurement_review", "family": "open_derived_missing_antecedent", "subset": "core_contrastive", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["has_risk_note", "has_vendor_approval", "may_receive_purchase_order", "needs_procurement_review", "submitted_tax_form"], "agent_a_complete_predicates": ["has_risk_note", "has_vendor_approval", "may_receive_purchase_order", "needs_procurement_review", "submitted_tax_form"], "agent_b_open_predicates": [], "facts_positive": ["Vendor A submitted a tax form."], "facts_negative": [], "rules_natural": ["If a vendor has a risk note, then that vendor needs procurement review."], "predicate_glossary": {"submitted_tax_form": "submitted a tax form", "has_vendor_approval": "has vendor approval", "has_risk_note": "has a risk note", "may_receive_purchase_order": "may receive a purchase order", "needs_procurement_review": "needs procurement review"}, "vocabulary_predicates": ["has_risk_note", "has_vendor_approval", "may_receive_purchase_order", "needs_procurement_review", "submitted_tax_form"], "query_atom": "needs_procurement_review::Vendor A", "target_statement": "Vendor A needs procurement review.", "gold_reason_type": "global_closed_world_underivable", "symbolic": {"positive_atoms": ["submitted_tax_form::Vendor A"], "negative_atoms": [], "rules": [{"antecedents": ["has_risk_note::Vendor A"], "conclusion": "needs_procurement_review::Vendor A", "text": "If a vendor has a risk note, then that vendor needs procurement review."}], "query_atom": "needs_procurement_review::Vendor A", "closure_atoms": ["submitted_tax_form::Vendor A"]}, "source_reports": {"agent_a": {"positive": ["Vendor A submitted a tax form."], "negative": []}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "closed_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_risk_note, has_vendor_approval, may_receive_purchase_order, needs_procurement_review, submitted_tax_form.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- has_risk_note: has a risk note\n- has_vendor_approval: has vendor approval\n- may_receive_purchase_order: may receive a purchase order\n- needs_procurement_review: needs procurement review\n- submitted_tax_form: submitted a tax form\n\nCoordinator policy rules:\n- If a vendor has a risk note, then that vendor needs procurement review.\n\nAgent A report:\nPositive facts:\n- Vendor A submitted a tax form.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: needs_procurement_review::Vendor A.\nTarget statement: Vendor A needs procurement review.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "procurement_review_0300_open_derived_missing_antecedent__lcwa__multi_agent", "source_id": "procurement_review_0300_open_derived_missing_antecedent__lcwa", "base_id": "procurement_review_0300_open_derived_missing_antecedent", "split": "train", "domain": "procurement_review", "family": "open_derived_missing_antecedent", "subset": "core_contrastive", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_vendor_approval", "may_receive_purchase_order"], "agent_a_complete_predicates": ["has_vendor_approval", "may_receive_purchase_order"], "agent_b_open_predicates": ["has_risk_note", "needs_procurement_review", "submitted_tax_form"], "facts_positive": ["Vendor A submitted a tax form."], "facts_negative": [], "rules_natural": ["If a vendor has a risk note, then that vendor needs procurement review."], "predicate_glossary": {"submitted_tax_form": "submitted a tax form", "has_vendor_approval": "has vendor approval", "has_risk_note": "has a risk note", "may_receive_purchase_order": "may receive a purchase order", "needs_procurement_review": "needs procurement review"}, "vocabulary_predicates": ["has_risk_note", "has_vendor_approval", "may_receive_purchase_order", "needs_procurement_review", "submitted_tax_form"], "query_atom": "needs_procurement_review::Vendor A", "target_statement": "Vendor A needs procurement review.", "gold_reason_type": "local_open_world_underivable", "symbolic": {"positive_atoms": ["submitted_tax_form::Vendor A"], "negative_atoms": [], "rules": [{"antecedents": ["has_risk_note::Vendor A"], "conclusion": "needs_procurement_review::Vendor A", "text": "If a vendor has a risk note, then that vendor needs procurement review."}], "query_atom": "needs_procurement_review::Vendor A", "closure_atoms": ["submitted_tax_form::Vendor A"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Vendor A submitted a tax form."], "negative": []}}, "gold_truth_value": "unknown", "gold_source_used": "agent_b", "gold_closure_handling": "open_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_vendor_approval, may_receive_purchase_order.\nAgent B open/incomplete predicates: has_risk_note, needs_procurement_review, submitted_tax_form.\n\nPredicate glossary:\n- has_risk_note: has a risk note\n- has_vendor_approval: has vendor approval\n- may_receive_purchase_order: may receive a purchase order\n- needs_procurement_review: needs procurement review\n- submitted_tax_form: submitted a tax form\n\nCoordinator policy rules:\n- If a vendor has a risk note, then that vendor needs procurement review.\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Vendor A submitted a tax form.\nNegative facts:\n- none\n\nTarget atom: needs_procurement_review::Vendor A.\nTarget statement: Vendor A needs procurement review.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "procurement_review_0300_open_derived_missing_antecedent__owa__multi_agent", "source_id": "procurement_review_0300_open_derived_missing_antecedent__owa", "base_id": "procurement_review_0300_open_derived_missing_antecedent", "split": "train", "domain": "procurement_review", "family": "open_derived_missing_antecedent", "subset": "core_contrastive", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["has_risk_note", "has_vendor_approval", "may_receive_purchase_order", "needs_procurement_review", "submitted_tax_form"], "facts_positive": ["Vendor A submitted a tax form."], "facts_negative": [], "rules_natural": ["If a vendor has a risk note, then that vendor needs procurement review."], "predicate_glossary": {"submitted_tax_form": "submitted a tax form", "has_vendor_approval": "has vendor approval", "has_risk_note": "has a risk note", "may_receive_purchase_order": "may receive a purchase order", "needs_procurement_review": "needs procurement review"}, "vocabulary_predicates": ["has_risk_note", "has_vendor_approval", "may_receive_purchase_order", "needs_procurement_review", "submitted_tax_form"], "query_atom": "needs_procurement_review::Vendor A", "target_statement": "Vendor A needs procurement review.", "gold_reason_type": "open_world_underivable", "symbolic": {"positive_atoms": ["submitted_tax_form::Vendor A"], "negative_atoms": [], "rules": [{"antecedents": ["has_risk_note::Vendor A"], "conclusion": "needs_procurement_review::Vendor A", "text": "If a vendor has a risk note, then that vendor needs procurement review."}], "query_atom": "needs_procurement_review::Vendor A", "closure_atoms": ["submitted_tax_form::Vendor A"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Vendor A submitted a tax form."], "negative": []}}, "gold_truth_value": "unknown", "gold_source_used": "agent_b", "gold_closure_handling": "open_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: has_risk_note, has_vendor_approval, may_receive_purchase_order, needs_procurement_review, submitted_tax_form.\n\nPredicate glossary:\n- has_risk_note: has a risk note\n- has_vendor_approval: has vendor approval\n- may_receive_purchase_order: may receive a purchase order\n- needs_procurement_review: needs procurement review\n- submitted_tax_form: submitted a tax form\n\nCoordinator policy rules:\n- If a vendor has a risk note, then that vendor needs procurement review.\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Vendor A submitted a tax form.\nNegative facts:\n- none\n\nTarget atom: needs_procurement_review::Vendor A.\nTarget statement: Vendor A needs procurement review.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "procurement_review_0303_open_derived_missing_antecedent__cwa__multi_agent", "source_id": "procurement_review_0303_open_derived_missing_antecedent__cwa", "base_id": "procurement_review_0303_open_derived_missing_antecedent", "split": "train", "domain": "procurement_review", "family": "open_derived_missing_antecedent", "subset": "core_contrastive", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["has_risk_note", "has_vendor_approval", "may_receive_purchase_order", "needs_procurement_review", "submitted_tax_form"], "agent_a_complete_predicates": ["has_risk_note", "has_vendor_approval", "may_receive_purchase_order", "needs_procurement_review", "submitted_tax_form"], "agent_b_open_predicates": [], "facts_positive": ["Vendor D submitted a tax form."], "facts_negative": [], "rules_natural": ["If a vendor has a risk note, then that vendor needs procurement review."], "predicate_glossary": {"submitted_tax_form": "submitted a tax form", "has_vendor_approval": "has vendor approval", "has_risk_note": "has a risk note", "may_receive_purchase_order": "may receive a purchase order", "needs_procurement_review": "needs procurement review"}, "vocabulary_predicates": ["has_risk_note", "has_vendor_approval", "may_receive_purchase_order", "needs_procurement_review", "submitted_tax_form"], "query_atom": "needs_procurement_review::Vendor D", "target_statement": "Vendor D needs procurement review.", "gold_reason_type": "global_closed_world_underivable", "symbolic": {"positive_atoms": ["submitted_tax_form::Vendor D"], "negative_atoms": [], "rules": [{"antecedents": ["has_risk_note::Vendor D"], "conclusion": "needs_procurement_review::Vendor D", "text": "If a vendor has a risk note, then that vendor needs procurement review."}], "query_atom": "needs_procurement_review::Vendor D", "closure_atoms": ["submitted_tax_form::Vendor D"]}, "source_reports": {"agent_a": {"positive": ["Vendor D submitted a tax form."], "negative": []}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "closed_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_risk_note, has_vendor_approval, may_receive_purchase_order, needs_procurement_review, submitted_tax_form.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- has_risk_note: has a risk note\n- has_vendor_approval: has vendor approval\n- may_receive_purchase_order: may receive a purchase order\n- needs_procurement_review: needs procurement review\n- submitted_tax_form: submitted a tax form\n\nCoordinator policy rules:\n- If a vendor has a risk note, then that vendor needs procurement review.\n\nAgent A report:\nPositive facts:\n- Vendor D submitted a tax form.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: needs_procurement_review::Vendor D.\nTarget statement: Vendor D needs procurement review.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "procurement_review_0303_open_derived_missing_antecedent__lcwa__multi_agent", "source_id": "procurement_review_0303_open_derived_missing_antecedent__lcwa", "base_id": "procurement_review_0303_open_derived_missing_antecedent", "split": "train", "domain": "procurement_review", "family": "open_derived_missing_antecedent", "subset": "core_contrastive", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_vendor_approval", "may_receive_purchase_order"], "agent_a_complete_predicates": ["has_vendor_approval", "may_receive_purchase_order"], "agent_b_open_predicates": ["has_risk_note", "needs_procurement_review", "submitted_tax_form"], "facts_positive": ["Vendor D submitted a tax form."], "facts_negative": [], "rules_natural": ["If a vendor has a risk note, then that vendor needs procurement review."], "predicate_glossary": {"submitted_tax_form": "submitted a tax form", "has_vendor_approval": "has vendor approval", "has_risk_note": "has a risk note", "may_receive_purchase_order": "may receive a purchase order", "needs_procurement_review": "needs procurement review"}, "vocabulary_predicates": ["has_risk_note", "has_vendor_approval", "may_receive_purchase_order", "needs_procurement_review", "submitted_tax_form"], "query_atom": "needs_procurement_review::Vendor D", "target_statement": "Vendor D needs procurement review.", "gold_reason_type": "local_open_world_underivable", "symbolic": {"positive_atoms": ["submitted_tax_form::Vendor D"], "negative_atoms": [], "rules": [{"antecedents": ["has_risk_note::Vendor D"], "conclusion": "needs_procurement_review::Vendor D", "text": "If a vendor has a risk note, then that vendor needs procurement review."}], "query_atom": "needs_procurement_review::Vendor D", "closure_atoms": ["submitted_tax_form::Vendor D"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Vendor D submitted a tax form."], "negative": []}}, "gold_truth_value": "unknown", "gold_source_used": "agent_b", "gold_closure_handling": "open_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_vendor_approval, may_receive_purchase_order.\nAgent B open/incomplete predicates: has_risk_note, needs_procurement_review, submitted_tax_form.\n\nPredicate glossary:\n- has_risk_note: has a risk note\n- has_vendor_approval: has vendor approval\n- may_receive_purchase_order: may receive a purchase order\n- needs_procurement_review: needs procurement review\n- submitted_tax_form: submitted a tax form\n\nCoordinator policy rules:\n- If a vendor has a risk note, then that vendor needs procurement review.\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Vendor D submitted a tax form.\nNegative facts:\n- none\n\nTarget atom: needs_procurement_review::Vendor D.\nTarget statement: Vendor D needs procurement review.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "procurement_review_0303_open_derived_missing_antecedent__owa__multi_agent", "source_id": "procurement_review_0303_open_derived_missing_antecedent__owa", "base_id": "procurement_review_0303_open_derived_missing_antecedent", "split": "train", "domain": "procurement_review", "family": "open_derived_missing_antecedent", "subset": "core_contrastive", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["has_risk_note", "has_vendor_approval", "may_receive_purchase_order", "needs_procurement_review", "submitted_tax_form"], "facts_positive": ["Vendor D submitted a tax form."], "facts_negative": [], "rules_natural": ["If a vendor has a risk note, then that vendor needs procurement review."], "predicate_glossary": {"submitted_tax_form": "submitted a tax form", "has_vendor_approval": "has vendor approval", "has_risk_note": "has a risk note", "may_receive_purchase_order": "may receive a purchase order", "needs_procurement_review": "needs procurement review"}, "vocabulary_predicates": ["has_risk_note", "has_vendor_approval", "may_receive_purchase_order", "needs_procurement_review", "submitted_tax_form"], "query_atom": "needs_procurement_review::Vendor D", "target_statement": "Vendor D needs procurement review.", "gold_reason_type": "open_world_underivable", "symbolic": {"positive_atoms": ["submitted_tax_form::Vendor D"], "negative_atoms": [], "rules": [{"antecedents": ["has_risk_note::Vendor D"], "conclusion": "needs_procurement_review::Vendor D", "text": "If a vendor has a risk note, then that vendor needs procurement review."}], "query_atom": "needs_procurement_review::Vendor D", "closure_atoms": ["submitted_tax_form::Vendor D"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Vendor D submitted a tax form."], "negative": []}}, "gold_truth_value": "unknown", "gold_source_used": "agent_b", "gold_closure_handling": "open_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: has_risk_note, has_vendor_approval, may_receive_purchase_order, needs_procurement_review, submitted_tax_form.\n\nPredicate glossary:\n- has_risk_note: has a risk note\n- has_vendor_approval: has vendor approval\n- may_receive_purchase_order: may receive a purchase order\n- needs_procurement_review: needs procurement review\n- submitted_tax_form: submitted a tax form\n\nCoordinator policy rules:\n- If a vendor has a risk note, then that vendor needs procurement review.\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Vendor D submitted a tax form.\nNegative facts:\n- none\n\nTarget atom: needs_procurement_review::Vendor D.\nTarget statement: Vendor D needs procurement review.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "procurement_review_0306_entailed_closed_conclusion__cwa__multi_agent", "source_id": "procurement_review_0306_entailed_closed_conclusion__cwa", "base_id": "procurement_review_0306_entailed_closed_conclusion", "split": "train", "domain": "procurement_review", "family": "entailed_closed_conclusion", "subset": "control_entailed", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["has_risk_note", "has_vendor_approval", "may_receive_purchase_order", "needs_procurement_review", "submitted_tax_form"], "agent_a_complete_predicates": ["has_risk_note", "has_vendor_approval", "may_receive_purchase_order", "needs_procurement_review", "submitted_tax_form"], "agent_b_open_predicates": [], "facts_positive": ["Vendor A has vendor approval.", "Vendor A submitted a tax form."], "facts_negative": [], "rules_natural": ["If a vendor submitted a tax form and has vendor approval, then that vendor may receive a purchase order."], "predicate_glossary": {"submitted_tax_form": "submitted a tax form", "has_vendor_approval": "has vendor approval", "has_risk_note": "has a risk note", "may_receive_purchase_order": "may receive a purchase order", "needs_procurement_review": "needs procurement review"}, "vocabulary_predicates": ["has_risk_note", "has_vendor_approval", "may_receive_purchase_order", "needs_procurement_review", "submitted_tax_form"], "query_atom": "may_receive_purchase_order::Vendor A", "target_statement": "Vendor A may receive a purchase order.", "gold_reason_type": "entailed_by_fact_or_rule", "symbolic": {"positive_atoms": ["has_vendor_approval::Vendor A", "submitted_tax_form::Vendor A"], "negative_atoms": [], "rules": [{"antecedents": ["submitted_tax_form::Vendor A", "has_vendor_approval::Vendor A"], "conclusion": "may_receive_purchase_order::Vendor A", "text": "If a vendor submitted a tax form and has vendor approval, then that vendor may receive a purchase order."}], "query_atom": "may_receive_purchase_order::Vendor A", "closure_atoms": ["has_vendor_approval::Vendor A", "may_receive_purchase_order::Vendor A", "submitted_tax_form::Vendor A"]}, "source_reports": {"agent_a": {"positive": ["Vendor A has vendor approval.", "Vendor A submitted a tax form."], "negative": []}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "true", "gold_source_used": "agent_a", "gold_closure_handling": "entailed_or_explicit", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_risk_note, has_vendor_approval, may_receive_purchase_order, needs_procurement_review, submitted_tax_form.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- has_risk_note: has a risk note\n- has_vendor_approval: has vendor approval\n- may_receive_purchase_order: may receive a purchase order\n- needs_procurement_review: needs procurement review\n- submitted_tax_form: submitted a tax form\n\nCoordinator policy rules:\n- If a vendor submitted a tax form and has vendor approval, then that vendor may receive a purchase order.\n\nAgent A report:\nPositive facts:\n- Vendor A has vendor approval.\n- Vendor A submitted a tax form.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: may_receive_purchase_order::Vendor A.\nTarget statement: Vendor A may receive a purchase order.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "procurement_review_0306_entailed_closed_conclusion__lcwa__multi_agent", "source_id": "procurement_review_0306_entailed_closed_conclusion__lcwa", "base_id": "procurement_review_0306_entailed_closed_conclusion", "split": "train", "domain": "procurement_review", "family": "entailed_closed_conclusion", "subset": "control_entailed", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_vendor_approval", "may_receive_purchase_order"], "agent_a_complete_predicates": ["has_vendor_approval", "may_receive_purchase_order"], "agent_b_open_predicates": ["has_risk_note", "needs_procurement_review", "submitted_tax_form"], "facts_positive": ["Vendor A has vendor approval.", "Vendor A submitted a tax form."], "facts_negative": [], "rules_natural": ["If a vendor submitted a tax form and has vendor approval, then that vendor may receive a purchase order."], "predicate_glossary": {"submitted_tax_form": "submitted a tax form", "has_vendor_approval": "has vendor approval", "has_risk_note": "has a risk note", "may_receive_purchase_order": "may receive a purchase order", "needs_procurement_review": "needs procurement review"}, "vocabulary_predicates": ["has_risk_note", "has_vendor_approval", "may_receive_purchase_order", "needs_procurement_review", "submitted_tax_form"], "query_atom": "may_receive_purchase_order::Vendor A", "target_statement": "Vendor A may receive a purchase order.", "gold_reason_type": "entailed_by_fact_or_rule", "symbolic": {"positive_atoms": ["has_vendor_approval::Vendor A", "submitted_tax_form::Vendor A"], "negative_atoms": [], "rules": [{"antecedents": ["submitted_tax_form::Vendor A", "has_vendor_approval::Vendor A"], "conclusion": "may_receive_purchase_order::Vendor A", "text": "If a vendor submitted a tax form and has vendor approval, then that vendor may receive a purchase order."}], "query_atom": "may_receive_purchase_order::Vendor A", "closure_atoms": ["has_vendor_approval::Vendor A", "may_receive_purchase_order::Vendor A", "submitted_tax_form::Vendor A"]}, "source_reports": {"agent_a": {"positive": ["Vendor A has vendor approval."], "negative": []}, "agent_b": {"positive": ["Vendor A submitted a tax form."], "negative": []}}, "gold_truth_value": "true", "gold_source_used": "both", "gold_closure_handling": "entailed_or_explicit", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_vendor_approval, may_receive_purchase_order.\nAgent B open/incomplete predicates: has_risk_note, needs_procurement_review, submitted_tax_form.\n\nPredicate glossary:\n- has_risk_note: has a risk note\n- has_vendor_approval: has vendor approval\n- may_receive_purchase_order: may receive a purchase order\n- needs_procurement_review: needs procurement review\n- submitted_tax_form: submitted a tax form\n\nCoordinator policy rules:\n- If a vendor submitted a tax form and has vendor approval, then that vendor may receive a purchase order.\n\nAgent A report:\nPositive facts:\n- Vendor A has vendor approval.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Vendor A submitted a tax form.\nNegative facts:\n- none\n\nTarget atom: may_receive_purchase_order::Vendor A.\nTarget statement: Vendor A may receive a purchase order.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "procurement_review_0306_entailed_closed_conclusion__owa__multi_agent", "source_id": "procurement_review_0306_entailed_closed_conclusion__owa", "base_id": "procurement_review_0306_entailed_closed_conclusion", "split": "train", "domain": "procurement_review", "family": "entailed_closed_conclusion", "subset": "control_entailed", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["has_risk_note", "has_vendor_approval", "may_receive_purchase_order", "needs_procurement_review", "submitted_tax_form"], "facts_positive": ["Vendor A has vendor approval.", "Vendor A submitted a tax form."], "facts_negative": [], "rules_natural": ["If a vendor submitted a tax form and has vendor approval, then that vendor may receive a purchase order."], "predicate_glossary": {"submitted_tax_form": "submitted a tax form", "has_vendor_approval": "has vendor approval", "has_risk_note": "has a risk note", "may_receive_purchase_order": "may receive a purchase order", "needs_procurement_review": "needs procurement review"}, "vocabulary_predicates": ["has_risk_note", "has_vendor_approval", "may_receive_purchase_order", "needs_procurement_review", "submitted_tax_form"], "query_atom": "may_receive_purchase_order::Vendor A", "target_statement": "Vendor A may receive a purchase order.", "gold_reason_type": "entailed_by_fact_or_rule", "symbolic": {"positive_atoms": ["has_vendor_approval::Vendor A", "submitted_tax_form::Vendor A"], "negative_atoms": [], "rules": [{"antecedents": ["submitted_tax_form::Vendor A", "has_vendor_approval::Vendor A"], "conclusion": "may_receive_purchase_order::Vendor A", "text": "If a vendor submitted a tax form and has vendor approval, then that vendor may receive a purchase order."}], "query_atom": "may_receive_purchase_order::Vendor A", "closure_atoms": ["has_vendor_approval::Vendor A", "may_receive_purchase_order::Vendor A", "submitted_tax_form::Vendor A"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Vendor A has vendor approval.", "Vendor A submitted a tax form."], "negative": []}}, "gold_truth_value": "true", "gold_source_used": "agent_b", "gold_closure_handling": "entailed_or_explicit", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: has_risk_note, has_vendor_approval, may_receive_purchase_order, needs_procurement_review, submitted_tax_form.\n\nPredicate glossary:\n- has_risk_note: has a risk note\n- has_vendor_approval: has vendor approval\n- may_receive_purchase_order: may receive a purchase order\n- needs_procurement_review: needs procurement review\n- submitted_tax_form: submitted a tax form\n\nCoordinator policy rules:\n- If a vendor submitted a tax form and has vendor approval, then that vendor may receive a purchase order.\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Vendor A has vendor approval.\n- Vendor A submitted a tax form.\nNegative facts:\n- none\n\nTarget atom: may_receive_purchase_order::Vendor A.\nTarget statement: Vendor A may receive a purchase order.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "procurement_review_0311_entailed_open_conclusion__cwa__multi_agent", "source_id": "procurement_review_0311_entailed_open_conclusion__cwa", "base_id": "procurement_review_0311_entailed_open_conclusion", "split": "train", "domain": "procurement_review", "family": "entailed_open_conclusion", "subset": "control_entailed", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["has_risk_note", "has_vendor_approval", "may_receive_purchase_order", "needs_procurement_review", "submitted_tax_form"], "agent_a_complete_predicates": ["has_risk_note", "has_vendor_approval", "may_receive_purchase_order", "needs_procurement_review", "submitted_tax_form"], "agent_b_open_predicates": [], "facts_positive": ["Vendor F has a risk note."], "facts_negative": [], "rules_natural": ["If a vendor has a risk note, then that vendor needs procurement review."], "predicate_glossary": {"submitted_tax_form": "submitted a tax form", "has_vendor_approval": "has vendor approval", "has_risk_note": "has a risk note", "may_receive_purchase_order": "may receive a purchase order", "needs_procurement_review": "needs procurement review"}, "vocabulary_predicates": ["has_risk_note", "has_vendor_approval", "may_receive_purchase_order", "needs_procurement_review", "submitted_tax_form"], "query_atom": "needs_procurement_review::Vendor F", "target_statement": "Vendor F needs procurement review.", "gold_reason_type": "entailed_by_fact_or_rule", "symbolic": {"positive_atoms": ["has_risk_note::Vendor F"], "negative_atoms": [], "rules": [{"antecedents": ["has_risk_note::Vendor F"], "conclusion": "needs_procurement_review::Vendor F", "text": "If a vendor has a risk note, then that vendor needs procurement review."}], "query_atom": "needs_procurement_review::Vendor F", "closure_atoms": ["has_risk_note::Vendor F", "needs_procurement_review::Vendor F"]}, "source_reports": {"agent_a": {"positive": ["Vendor F has a risk note."], "negative": []}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "true", "gold_source_used": "agent_a", "gold_closure_handling": "entailed_or_explicit", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_risk_note, has_vendor_approval, may_receive_purchase_order, needs_procurement_review, submitted_tax_form.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- has_risk_note: has a risk note\n- has_vendor_approval: has vendor approval\n- may_receive_purchase_order: may receive a purchase order\n- needs_procurement_review: needs procurement review\n- submitted_tax_form: submitted a tax form\n\nCoordinator policy rules:\n- If a vendor has a risk note, then that vendor needs procurement review.\n\nAgent A report:\nPositive facts:\n- Vendor F has a risk note.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: needs_procurement_review::Vendor F.\nTarget statement: Vendor F needs procurement review.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "procurement_review_0311_entailed_open_conclusion__lcwa__multi_agent", "source_id": "procurement_review_0311_entailed_open_conclusion__lcwa", "base_id": "procurement_review_0311_entailed_open_conclusion", "split": "train", "domain": "procurement_review", "family": "entailed_open_conclusion", "subset": "control_entailed", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_vendor_approval", "may_receive_purchase_order"], "agent_a_complete_predicates": ["has_vendor_approval", "may_receive_purchase_order"], "agent_b_open_predicates": ["has_risk_note", "needs_procurement_review", "submitted_tax_form"], "facts_positive": ["Vendor F has a risk note."], "facts_negative": [], "rules_natural": ["If a vendor has a risk note, then that vendor needs procurement review."], "predicate_glossary": {"submitted_tax_form": "submitted a tax form", "has_vendor_approval": "has vendor approval", "has_risk_note": "has a risk note", "may_receive_purchase_order": "may receive a purchase order", "needs_procurement_review": "needs procurement review"}, "vocabulary_predicates": ["has_risk_note", "has_vendor_approval", "may_receive_purchase_order", "needs_procurement_review", "submitted_tax_form"], "query_atom": "needs_procurement_review::Vendor F", "target_statement": "Vendor F needs procurement review.", "gold_reason_type": "entailed_by_fact_or_rule", "symbolic": {"positive_atoms": ["has_risk_note::Vendor F"], "negative_atoms": [], "rules": [{"antecedents": ["has_risk_note::Vendor F"], "conclusion": "needs_procurement_review::Vendor F", "text": "If a vendor has a risk note, then that vendor needs procurement review."}], "query_atom": "needs_procurement_review::Vendor F", "closure_atoms": ["has_risk_note::Vendor F", "needs_procurement_review::Vendor F"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Vendor F has a risk note."], "negative": []}}, "gold_truth_value": "true", "gold_source_used": "agent_b", "gold_closure_handling": "entailed_or_explicit", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_vendor_approval, may_receive_purchase_order.\nAgent B open/incomplete predicates: has_risk_note, needs_procurement_review, submitted_tax_form.\n\nPredicate glossary:\n- has_risk_note: has a risk note\n- has_vendor_approval: has vendor approval\n- may_receive_purchase_order: may receive a purchase order\n- needs_procurement_review: needs procurement review\n- submitted_tax_form: submitted a tax form\n\nCoordinator policy rules:\n- If a vendor has a risk note, then that vendor needs procurement review.\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Vendor F has a risk note.\nNegative facts:\n- none\n\nTarget atom: needs_procurement_review::Vendor F.\nTarget statement: Vendor F needs procurement review.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "procurement_review_0311_entailed_open_conclusion__owa__multi_agent", "source_id": "procurement_review_0311_entailed_open_conclusion__owa", "base_id": "procurement_review_0311_entailed_open_conclusion", "split": "train", "domain": "procurement_review", "family": "entailed_open_conclusion", "subset": "control_entailed", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["has_risk_note", "has_vendor_approval", "may_receive_purchase_order", "needs_procurement_review", "submitted_tax_form"], "facts_positive": ["Vendor F has a risk note."], "facts_negative": [], "rules_natural": ["If a vendor has a risk note, then that vendor needs procurement review."], "predicate_glossary": {"submitted_tax_form": "submitted a tax form", "has_vendor_approval": "has vendor approval", "has_risk_note": "has a risk note", "may_receive_purchase_order": "may receive a purchase order", "needs_procurement_review": "needs procurement review"}, "vocabulary_predicates": ["has_risk_note", "has_vendor_approval", "may_receive_purchase_order", "needs_procurement_review", "submitted_tax_form"], "query_atom": "needs_procurement_review::Vendor F", "target_statement": "Vendor F needs procurement review.", "gold_reason_type": "entailed_by_fact_or_rule", "symbolic": {"positive_atoms": ["has_risk_note::Vendor F"], "negative_atoms": [], "rules": [{"antecedents": ["has_risk_note::Vendor F"], "conclusion": "needs_procurement_review::Vendor F", "text": "If a vendor has a risk note, then that vendor needs procurement review."}], "query_atom": "needs_procurement_review::Vendor F", "closure_atoms": ["has_risk_note::Vendor F", "needs_procurement_review::Vendor F"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Vendor F has a risk note."], "negative": []}}, "gold_truth_value": "true", "gold_source_used": "agent_b", "gold_closure_handling": "entailed_or_explicit", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: has_risk_note, has_vendor_approval, may_receive_purchase_order, needs_procurement_review, submitted_tax_form.\n\nPredicate glossary:\n- has_risk_note: has a risk note\n- has_vendor_approval: has vendor approval\n- may_receive_purchase_order: may receive a purchase order\n- needs_procurement_review: needs procurement review\n- submitted_tax_form: submitted a tax form\n\nCoordinator policy rules:\n- If a vendor has a risk note, then that vendor needs procurement review.\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Vendor F has a risk note.\nNegative facts:\n- none\n\nTarget atom: needs_procurement_review::Vendor F.\nTarget statement: Vendor F needs procurement review.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "procurement_review_0313_explicit_negative_closed__cwa__multi_agent", "source_id": "procurement_review_0313_explicit_negative_closed__cwa", "base_id": "procurement_review_0313_explicit_negative_closed", "split": "test", "domain": "procurement_review", "family": "explicit_negative_closed", "subset": "control_explicit_negative", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["has_risk_note", "has_vendor_approval", "may_receive_purchase_order", "needs_procurement_review", "submitted_tax_form"], "agent_a_complete_predicates": ["has_risk_note", "has_vendor_approval", "may_receive_purchase_order", "needs_procurement_review", "submitted_tax_form"], "agent_b_open_predicates": [], "facts_positive": ["Vendor D submitted a tax form."], "facts_negative": ["Vendor B does not has vendor approval."], "rules_natural": [], "predicate_glossary": {"submitted_tax_form": "submitted a tax form", "has_vendor_approval": "has vendor approval", "has_risk_note": "has a risk note", "may_receive_purchase_order": "may receive a purchase order", "needs_procurement_review": "needs procurement review"}, "vocabulary_predicates": ["has_risk_note", "has_vendor_approval", "may_receive_purchase_order", "needs_procurement_review", "submitted_tax_form"], "query_atom": "has_vendor_approval::Vendor B", "target_statement": "Vendor B has vendor approval.", "gold_reason_type": "explicit_negative_fact", "symbolic": {"positive_atoms": ["submitted_tax_form::Vendor D"], "negative_atoms": ["has_vendor_approval::Vendor B"], "rules": [], "query_atom": "has_vendor_approval::Vendor B", "closure_atoms": ["submitted_tax_form::Vendor D"]}, "source_reports": {"agent_a": {"positive": ["Vendor D submitted a tax form."], "negative": ["Vendor B does not has vendor approval."]}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "explicit_negative", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_risk_note, has_vendor_approval, may_receive_purchase_order, needs_procurement_review, submitted_tax_form.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- has_risk_note: has a risk note\n- has_vendor_approval: has vendor approval\n- may_receive_purchase_order: may receive a purchase order\n- needs_procurement_review: needs procurement review\n- submitted_tax_form: submitted a tax form\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- Vendor D submitted a tax form.\nNegative facts:\n- Vendor B does not has vendor approval.\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: has_vendor_approval::Vendor B.\nTarget statement: Vendor B has vendor approval.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "procurement_review_0313_explicit_negative_closed__lcwa__multi_agent", "source_id": "procurement_review_0313_explicit_negative_closed__lcwa", "base_id": "procurement_review_0313_explicit_negative_closed", "split": "test", "domain": "procurement_review", "family": "explicit_negative_closed", "subset": "control_explicit_negative", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_vendor_approval"], "agent_a_complete_predicates": ["has_vendor_approval"], "agent_b_open_predicates": ["has_risk_note", "may_receive_purchase_order", "needs_procurement_review", "submitted_tax_form"], "facts_positive": ["Vendor D submitted a tax form."], "facts_negative": ["Vendor B does not has vendor approval."], "rules_natural": [], "predicate_glossary": {"submitted_tax_form": "submitted a tax form", "has_vendor_approval": "has vendor approval", "has_risk_note": "has a risk note", "may_receive_purchase_order": "may receive a purchase order", "needs_procurement_review": "needs procurement review"}, "vocabulary_predicates": ["has_risk_note", "has_vendor_approval", "may_receive_purchase_order", "needs_procurement_review", "submitted_tax_form"], "query_atom": "has_vendor_approval::Vendor B", "target_statement": "Vendor B has vendor approval.", "gold_reason_type": "explicit_negative_fact", "symbolic": {"positive_atoms": ["submitted_tax_form::Vendor D"], "negative_atoms": ["has_vendor_approval::Vendor B"], "rules": [], "query_atom": "has_vendor_approval::Vendor B", "closure_atoms": ["submitted_tax_form::Vendor D"]}, "source_reports": {"agent_a": {"positive": [], "negative": ["Vendor B does not has vendor approval."]}, "agent_b": {"positive": ["Vendor D submitted a tax form."], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "explicit_negative", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_vendor_approval.\nAgent B open/incomplete predicates: has_risk_note, may_receive_purchase_order, needs_procurement_review, submitted_tax_form.\n\nPredicate glossary:\n- has_risk_note: has a risk note\n- has_vendor_approval: has vendor approval\n- may_receive_purchase_order: may receive a purchase order\n- needs_procurement_review: needs procurement review\n- submitted_tax_form: submitted a tax form\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- Vendor B does not has vendor approval.\n\nAgent B report:\nPositive facts:\n- Vendor D submitted a tax form.\nNegative facts:\n- none\n\nTarget atom: has_vendor_approval::Vendor B.\nTarget statement: Vendor B has vendor approval.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "procurement_review_0313_explicit_negative_closed__owa__multi_agent", "source_id": "procurement_review_0313_explicit_negative_closed__owa", "base_id": "procurement_review_0313_explicit_negative_closed", "split": "test", "domain": "procurement_review", "family": "explicit_negative_closed", "subset": "control_explicit_negative", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["has_risk_note", "has_vendor_approval", "may_receive_purchase_order", "needs_procurement_review", "submitted_tax_form"], "facts_positive": ["Vendor D submitted a tax form."], "facts_negative": ["Vendor B does not has vendor approval."], "rules_natural": [], "predicate_glossary": {"submitted_tax_form": "submitted a tax form", "has_vendor_approval": "has vendor approval", "has_risk_note": "has a risk note", "may_receive_purchase_order": "may receive a purchase order", "needs_procurement_review": "needs procurement review"}, "vocabulary_predicates": ["has_risk_note", "has_vendor_approval", "may_receive_purchase_order", "needs_procurement_review", "submitted_tax_form"], "query_atom": "has_vendor_approval::Vendor B", "target_statement": "Vendor B has vendor approval.", "gold_reason_type": "explicit_negative_fact", "symbolic": {"positive_atoms": ["submitted_tax_form::Vendor D"], "negative_atoms": ["has_vendor_approval::Vendor B"], "rules": [], "query_atom": "has_vendor_approval::Vendor B", "closure_atoms": ["submitted_tax_form::Vendor D"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Vendor D submitted a tax form."], "negative": ["Vendor B does not has vendor approval."]}}, "gold_truth_value": "false", "gold_source_used": "agent_b", "gold_closure_handling": "explicit_negative", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: has_risk_note, has_vendor_approval, may_receive_purchase_order, needs_procurement_review, submitted_tax_form.\n\nPredicate glossary:\n- has_risk_note: has a risk note\n- has_vendor_approval: has vendor approval\n- may_receive_purchase_order: may receive a purchase order\n- needs_procurement_review: needs procurement review\n- submitted_tax_form: submitted a tax form\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Vendor D submitted a tax form.\nNegative facts:\n- Vendor B does not has vendor approval.\n\nTarget atom: has_vendor_approval::Vendor B.\nTarget statement: Vendor B has vendor approval.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "procurement_review_0315_explicit_negative_closed__cwa__multi_agent", "source_id": "procurement_review_0315_explicit_negative_closed__cwa", "base_id": "procurement_review_0315_explicit_negative_closed", "split": "test", "domain": "procurement_review", "family": "explicit_negative_closed", "subset": "control_explicit_negative", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["has_risk_note", "has_vendor_approval", "may_receive_purchase_order", "needs_procurement_review", "submitted_tax_form"], "agent_a_complete_predicates": ["has_risk_note", "has_vendor_approval", "may_receive_purchase_order", "needs_procurement_review", "submitted_tax_form"], "agent_b_open_predicates": [], "facts_positive": ["Vendor F submitted a tax form."], "facts_negative": ["Vendor D does not has vendor approval."], "rules_natural": [], "predicate_glossary": {"submitted_tax_form": "submitted a tax form", "has_vendor_approval": "has vendor approval", "has_risk_note": "has a risk note", "may_receive_purchase_order": "may receive a purchase order", "needs_procurement_review": "needs procurement review"}, "vocabulary_predicates": ["has_risk_note", "has_vendor_approval", "may_receive_purchase_order", "needs_procurement_review", "submitted_tax_form"], "query_atom": "has_vendor_approval::Vendor D", "target_statement": "Vendor D has vendor approval.", "gold_reason_type": "explicit_negative_fact", "symbolic": {"positive_atoms": ["submitted_tax_form::Vendor F"], "negative_atoms": ["has_vendor_approval::Vendor D"], "rules": [], "query_atom": "has_vendor_approval::Vendor D", "closure_atoms": ["submitted_tax_form::Vendor F"]}, "source_reports": {"agent_a": {"positive": ["Vendor F submitted a tax form."], "negative": ["Vendor D does not has vendor approval."]}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "explicit_negative", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_risk_note, has_vendor_approval, may_receive_purchase_order, needs_procurement_review, submitted_tax_form.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- has_risk_note: has a risk note\n- has_vendor_approval: has vendor approval\n- may_receive_purchase_order: may receive a purchase order\n- needs_procurement_review: needs procurement review\n- submitted_tax_form: submitted a tax form\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- Vendor F submitted a tax form.\nNegative facts:\n- Vendor D does not has vendor approval.\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: has_vendor_approval::Vendor D.\nTarget statement: Vendor D has vendor approval.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "procurement_review_0315_explicit_negative_closed__lcwa__multi_agent", "source_id": "procurement_review_0315_explicit_negative_closed__lcwa", "base_id": "procurement_review_0315_explicit_negative_closed", "split": "test", "domain": "procurement_review", "family": "explicit_negative_closed", "subset": "control_explicit_negative", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_vendor_approval"], "agent_a_complete_predicates": ["has_vendor_approval"], "agent_b_open_predicates": ["has_risk_note", "may_receive_purchase_order", "needs_procurement_review", "submitted_tax_form"], "facts_positive": ["Vendor F submitted a tax form."], "facts_negative": ["Vendor D does not has vendor approval."], "rules_natural": [], "predicate_glossary": {"submitted_tax_form": "submitted a tax form", "has_vendor_approval": "has vendor approval", "has_risk_note": "has a risk note", "may_receive_purchase_order": "may receive a purchase order", "needs_procurement_review": "needs procurement review"}, "vocabulary_predicates": ["has_risk_note", "has_vendor_approval", "may_receive_purchase_order", "needs_procurement_review", "submitted_tax_form"], "query_atom": "has_vendor_approval::Vendor D", "target_statement": "Vendor D has vendor approval.", "gold_reason_type": "explicit_negative_fact", "symbolic": {"positive_atoms": ["submitted_tax_form::Vendor F"], "negative_atoms": ["has_vendor_approval::Vendor D"], "rules": [], "query_atom": "has_vendor_approval::Vendor D", "closure_atoms": ["submitted_tax_form::Vendor F"]}, "source_reports": {"agent_a": {"positive": [], "negative": ["Vendor D does not has vendor approval."]}, "agent_b": {"positive": ["Vendor F submitted a tax form."], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "explicit_negative", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_vendor_approval.\nAgent B open/incomplete predicates: has_risk_note, may_receive_purchase_order, needs_procurement_review, submitted_tax_form.\n\nPredicate glossary:\n- has_risk_note: has a risk note\n- has_vendor_approval: has vendor approval\n- may_receive_purchase_order: may receive a purchase order\n- needs_procurement_review: needs procurement review\n- submitted_tax_form: submitted a tax form\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- Vendor D does not has vendor approval.\n\nAgent B report:\nPositive facts:\n- Vendor F submitted a tax form.\nNegative facts:\n- none\n\nTarget atom: has_vendor_approval::Vendor D.\nTarget statement: Vendor D has vendor approval.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "procurement_review_0315_explicit_negative_closed__owa__multi_agent", "source_id": "procurement_review_0315_explicit_negative_closed__owa", "base_id": "procurement_review_0315_explicit_negative_closed", "split": "test", "domain": "procurement_review", "family": "explicit_negative_closed", "subset": "control_explicit_negative", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["has_risk_note", "has_vendor_approval", "may_receive_purchase_order", "needs_procurement_review", "submitted_tax_form"], "facts_positive": ["Vendor F submitted a tax form."], "facts_negative": ["Vendor D does not has vendor approval."], "rules_natural": [], "predicate_glossary": {"submitted_tax_form": "submitted a tax form", "has_vendor_approval": "has vendor approval", "has_risk_note": "has a risk note", "may_receive_purchase_order": "may receive a purchase order", "needs_procurement_review": "needs procurement review"}, "vocabulary_predicates": ["has_risk_note", "has_vendor_approval", "may_receive_purchase_order", "needs_procurement_review", "submitted_tax_form"], "query_atom": "has_vendor_approval::Vendor D", "target_statement": "Vendor D has vendor approval.", "gold_reason_type": "explicit_negative_fact", "symbolic": {"positive_atoms": ["submitted_tax_form::Vendor F"], "negative_atoms": ["has_vendor_approval::Vendor D"], "rules": [], "query_atom": "has_vendor_approval::Vendor D", "closure_atoms": ["submitted_tax_form::Vendor F"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Vendor F submitted a tax form."], "negative": ["Vendor D does not has vendor approval."]}}, "gold_truth_value": "false", "gold_source_used": "agent_b", "gold_closure_handling": "explicit_negative", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: has_risk_note, has_vendor_approval, may_receive_purchase_order, needs_procurement_review, submitted_tax_form.\n\nPredicate glossary:\n- has_risk_note: has a risk note\n- has_vendor_approval: has vendor approval\n- may_receive_purchase_order: may receive a purchase order\n- needs_procurement_review: needs procurement review\n- submitted_tax_form: submitted a tax form\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Vendor F submitted a tax form.\nNegative facts:\n- Vendor D does not has vendor approval.\n\nTarget atom: has_vendor_approval::Vendor D.\nTarget statement: Vendor D has vendor approval.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "procurement_review_0316_explicit_negative_open__cwa__multi_agent", "source_id": "procurement_review_0316_explicit_negative_open__cwa", "base_id": "procurement_review_0316_explicit_negative_open", "split": "test", "domain": "procurement_review", "family": "explicit_negative_open", "subset": "control_explicit_negative", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["has_risk_note", "has_vendor_approval", "may_receive_purchase_order", "needs_procurement_review", "submitted_tax_form"], "agent_a_complete_predicates": ["has_risk_note", "has_vendor_approval", "may_receive_purchase_order", "needs_procurement_review", "submitted_tax_form"], "agent_b_open_predicates": [], "facts_positive": ["Vendor A submitted a tax form."], "facts_negative": ["Vendor E does not has a risk note."], "rules_natural": [], "predicate_glossary": {"submitted_tax_form": "submitted a tax form", "has_vendor_approval": "has vendor approval", "has_risk_note": "has a risk note", "may_receive_purchase_order": "may receive a purchase order", "needs_procurement_review": "needs procurement review"}, "vocabulary_predicates": ["has_risk_note", "has_vendor_approval", "may_receive_purchase_order", "needs_procurement_review", "submitted_tax_form"], "query_atom": "has_risk_note::Vendor E", "target_statement": "Vendor E has a risk note.", "gold_reason_type": "explicit_negative_fact", "symbolic": {"positive_atoms": ["submitted_tax_form::Vendor A"], "negative_atoms": ["has_risk_note::Vendor E"], "rules": [], "query_atom": "has_risk_note::Vendor E", "closure_atoms": ["submitted_tax_form::Vendor A"]}, "source_reports": {"agent_a": {"positive": ["Vendor A submitted a tax form."], "negative": ["Vendor E does not has a risk note."]}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "explicit_negative", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_risk_note, has_vendor_approval, may_receive_purchase_order, needs_procurement_review, submitted_tax_form.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- has_risk_note: has a risk note\n- has_vendor_approval: has vendor approval\n- may_receive_purchase_order: may receive a purchase order\n- needs_procurement_review: needs procurement review\n- submitted_tax_form: submitted a tax form\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- Vendor A submitted a tax form.\nNegative facts:\n- Vendor E does not has a risk note.\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: has_risk_note::Vendor E.\nTarget statement: Vendor E has a risk note.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "procurement_review_0316_explicit_negative_open__lcwa__multi_agent", "source_id": "procurement_review_0316_explicit_negative_open__lcwa", "base_id": "procurement_review_0316_explicit_negative_open", "split": "test", "domain": "procurement_review", "family": "explicit_negative_open", "subset": "control_explicit_negative", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_vendor_approval", "may_receive_purchase_order"], "agent_a_complete_predicates": ["has_vendor_approval", "may_receive_purchase_order"], "agent_b_open_predicates": ["has_risk_note", "needs_procurement_review", "submitted_tax_form"], "facts_positive": ["Vendor A submitted a tax form."], "facts_negative": ["Vendor E does not has a risk note."], "rules_natural": [], "predicate_glossary": {"submitted_tax_form": "submitted a tax form", "has_vendor_approval": "has vendor approval", "has_risk_note": "has a risk note", "may_receive_purchase_order": "may receive a purchase order", "needs_procurement_review": "needs procurement review"}, "vocabulary_predicates": ["has_risk_note", "has_vendor_approval", "may_receive_purchase_order", "needs_procurement_review", "submitted_tax_form"], "query_atom": "has_risk_note::Vendor E", "target_statement": "Vendor E has a risk note.", "gold_reason_type": "explicit_negative_fact", "symbolic": {"positive_atoms": ["submitted_tax_form::Vendor A"], "negative_atoms": ["has_risk_note::Vendor E"], "rules": [], "query_atom": "has_risk_note::Vendor E", "closure_atoms": ["submitted_tax_form::Vendor A"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Vendor A submitted a tax form."], "negative": ["Vendor E does not has a risk note."]}}, "gold_truth_value": "false", "gold_source_used": "agent_b", "gold_closure_handling": "explicit_negative", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_vendor_approval, may_receive_purchase_order.\nAgent B open/incomplete predicates: has_risk_note, needs_procurement_review, submitted_tax_form.\n\nPredicate glossary:\n- has_risk_note: has a risk note\n- has_vendor_approval: has vendor approval\n- may_receive_purchase_order: may receive a purchase order\n- needs_procurement_review: needs procurement review\n- submitted_tax_form: submitted a tax form\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Vendor A submitted a tax form.\nNegative facts:\n- Vendor E does not has a risk note.\n\nTarget atom: has_risk_note::Vendor E.\nTarget statement: Vendor E has a risk note.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "procurement_review_0316_explicit_negative_open__owa__multi_agent", "source_id": "procurement_review_0316_explicit_negative_open__owa", "base_id": "procurement_review_0316_explicit_negative_open", "split": "test", "domain": "procurement_review", "family": "explicit_negative_open", "subset": "control_explicit_negative", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["has_risk_note", "has_vendor_approval", "may_receive_purchase_order", "needs_procurement_review", "submitted_tax_form"], "facts_positive": ["Vendor A submitted a tax form."], "facts_negative": ["Vendor E does not has a risk note."], "rules_natural": [], "predicate_glossary": {"submitted_tax_form": "submitted a tax form", "has_vendor_approval": "has vendor approval", "has_risk_note": "has a risk note", "may_receive_purchase_order": "may receive a purchase order", "needs_procurement_review": "needs procurement review"}, "vocabulary_predicates": ["has_risk_note", "has_vendor_approval", "may_receive_purchase_order", "needs_procurement_review", "submitted_tax_form"], "query_atom": "has_risk_note::Vendor E", "target_statement": "Vendor E has a risk note.", "gold_reason_type": "explicit_negative_fact", "symbolic": {"positive_atoms": ["submitted_tax_form::Vendor A"], "negative_atoms": ["has_risk_note::Vendor E"], "rules": [], "query_atom": "has_risk_note::Vendor E", "closure_atoms": ["submitted_tax_form::Vendor A"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Vendor A submitted a tax form."], "negative": ["Vendor E does not has a risk note."]}}, "gold_truth_value": "false", "gold_source_used": "agent_b", "gold_closure_handling": "explicit_negative", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: has_risk_note, has_vendor_approval, may_receive_purchase_order, needs_procurement_review, submitted_tax_form.\n\nPredicate glossary:\n- has_risk_note: has a risk note\n- has_vendor_approval: has vendor approval\n- may_receive_purchase_order: may receive a purchase order\n- needs_procurement_review: needs procurement review\n- submitted_tax_form: submitted a tax form\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Vendor A submitted a tax form.\nNegative facts:\n- Vendor E does not has a risk note.\n\nTarget atom: has_risk_note::Vendor E.\nTarget statement: Vendor E has a risk note.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "procurement_review_0317_explicit_negative_open__cwa__multi_agent", "source_id": "procurement_review_0317_explicit_negative_open__cwa", "base_id": "procurement_review_0317_explicit_negative_open", "split": "train", "domain": "procurement_review", "family": "explicit_negative_open", "subset": "control_explicit_negative", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["has_risk_note", "has_vendor_approval", "may_receive_purchase_order", "needs_procurement_review", "submitted_tax_form"], "agent_a_complete_predicates": ["has_risk_note", "has_vendor_approval", "may_receive_purchase_order", "needs_procurement_review", "submitted_tax_form"], "agent_b_open_predicates": [], "facts_positive": ["Vendor B submitted a tax form."], "facts_negative": ["Vendor F does not has a risk note."], "rules_natural": [], "predicate_glossary": {"submitted_tax_form": "submitted a tax form", "has_vendor_approval": "has vendor approval", "has_risk_note": "has a risk note", "may_receive_purchase_order": "may receive a purchase order", "needs_procurement_review": "needs procurement review"}, "vocabulary_predicates": ["has_risk_note", "has_vendor_approval", "may_receive_purchase_order", "needs_procurement_review", "submitted_tax_form"], "query_atom": "has_risk_note::Vendor F", "target_statement": "Vendor F has a risk note.", "gold_reason_type": "explicit_negative_fact", "symbolic": {"positive_atoms": ["submitted_tax_form::Vendor B"], "negative_atoms": ["has_risk_note::Vendor F"], "rules": [], "query_atom": "has_risk_note::Vendor F", "closure_atoms": ["submitted_tax_form::Vendor B"]}, "source_reports": {"agent_a": {"positive": ["Vendor B submitted a tax form."], "negative": ["Vendor F does not has a risk note."]}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "explicit_negative", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_risk_note, has_vendor_approval, may_receive_purchase_order, needs_procurement_review, submitted_tax_form.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- has_risk_note: has a risk note\n- has_vendor_approval: has vendor approval\n- may_receive_purchase_order: may receive a purchase order\n- needs_procurement_review: needs procurement review\n- submitted_tax_form: submitted a tax form\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- Vendor B submitted a tax form.\nNegative facts:\n- Vendor F does not has a risk note.\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: has_risk_note::Vendor F.\nTarget statement: Vendor F has a risk note.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "procurement_review_0317_explicit_negative_open__lcwa__multi_agent", "source_id": "procurement_review_0317_explicit_negative_open__lcwa", "base_id": "procurement_review_0317_explicit_negative_open", "split": "train", "domain": "procurement_review", "family": "explicit_negative_open", "subset": "control_explicit_negative", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_vendor_approval", "may_receive_purchase_order"], "agent_a_complete_predicates": ["has_vendor_approval", "may_receive_purchase_order"], "agent_b_open_predicates": ["has_risk_note", "needs_procurement_review", "submitted_tax_form"], "facts_positive": ["Vendor B submitted a tax form."], "facts_negative": ["Vendor F does not has a risk note."], "rules_natural": [], "predicate_glossary": {"submitted_tax_form": "submitted a tax form", "has_vendor_approval": "has vendor approval", "has_risk_note": "has a risk note", "may_receive_purchase_order": "may receive a purchase order", "needs_procurement_review": "needs procurement review"}, "vocabulary_predicates": ["has_risk_note", "has_vendor_approval", "may_receive_purchase_order", "needs_procurement_review", "submitted_tax_form"], "query_atom": "has_risk_note::Vendor F", "target_statement": "Vendor F has a risk note.", "gold_reason_type": "explicit_negative_fact", "symbolic": {"positive_atoms": ["submitted_tax_form::Vendor B"], "negative_atoms": ["has_risk_note::Vendor F"], "rules": [], "query_atom": "has_risk_note::Vendor F", "closure_atoms": ["submitted_tax_form::Vendor B"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Vendor B submitted a tax form."], "negative": ["Vendor F does not has a risk note."]}}, "gold_truth_value": "false", "gold_source_used": "agent_b", "gold_closure_handling": "explicit_negative", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_vendor_approval, may_receive_purchase_order.\nAgent B open/incomplete predicates: has_risk_note, needs_procurement_review, submitted_tax_form.\n\nPredicate glossary:\n- has_risk_note: has a risk note\n- has_vendor_approval: has vendor approval\n- may_receive_purchase_order: may receive a purchase order\n- needs_procurement_review: needs procurement review\n- submitted_tax_form: submitted a tax form\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Vendor B submitted a tax form.\nNegative facts:\n- Vendor F does not has a risk note.\n\nTarget atom: has_risk_note::Vendor F.\nTarget statement: Vendor F has a risk note.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "procurement_review_0317_explicit_negative_open__owa__multi_agent", "source_id": "procurement_review_0317_explicit_negative_open__owa", "base_id": "procurement_review_0317_explicit_negative_open", "split": "train", "domain": "procurement_review", "family": "explicit_negative_open", "subset": "control_explicit_negative", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["has_risk_note", "has_vendor_approval", "may_receive_purchase_order", "needs_procurement_review", "submitted_tax_form"], "facts_positive": ["Vendor B submitted a tax form."], "facts_negative": ["Vendor F does not has a risk note."], "rules_natural": [], "predicate_glossary": {"submitted_tax_form": "submitted a tax form", "has_vendor_approval": "has vendor approval", "has_risk_note": "has a risk note", "may_receive_purchase_order": "may receive a purchase order", "needs_procurement_review": "needs procurement review"}, "vocabulary_predicates": ["has_risk_note", "has_vendor_approval", "may_receive_purchase_order", "needs_procurement_review", "submitted_tax_form"], "query_atom": "has_risk_note::Vendor F", "target_statement": "Vendor F has a risk note.", "gold_reason_type": "explicit_negative_fact", "symbolic": {"positive_atoms": ["submitted_tax_form::Vendor B"], "negative_atoms": ["has_risk_note::Vendor F"], "rules": [], "query_atom": "has_risk_note::Vendor F", "closure_atoms": ["submitted_tax_form::Vendor B"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Vendor B submitted a tax form."], "negative": ["Vendor F does not has a risk note."]}}, "gold_truth_value": "false", "gold_source_used": "agent_b", "gold_closure_handling": "explicit_negative", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: has_risk_note, has_vendor_approval, may_receive_purchase_order, needs_procurement_review, submitted_tax_form.\n\nPredicate glossary:\n- has_risk_note: has a risk note\n- has_vendor_approval: has vendor approval\n- may_receive_purchase_order: may receive a purchase order\n- needs_procurement_review: needs procurement review\n- submitted_tax_form: submitted a tax form\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Vendor B submitted a tax form.\nNegative facts:\n- Vendor F does not has a risk note.\n\nTarget atom: has_risk_note::Vendor F.\nTarget statement: Vendor F has a risk note.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "procurement_review_0318_explicit_negative_open__cwa__multi_agent", "source_id": "procurement_review_0318_explicit_negative_open__cwa", "base_id": "procurement_review_0318_explicit_negative_open", "split": "dev", "domain": "procurement_review", "family": "explicit_negative_open", "subset": "control_explicit_negative", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["has_risk_note", "has_vendor_approval", "may_receive_purchase_order", "needs_procurement_review", "submitted_tax_form"], "agent_a_complete_predicates": ["has_risk_note", "has_vendor_approval", "may_receive_purchase_order", "needs_procurement_review", "submitted_tax_form"], "agent_b_open_predicates": [], "facts_positive": ["Vendor C submitted a tax form."], "facts_negative": ["Vendor A does not has a risk note."], "rules_natural": [], "predicate_glossary": {"submitted_tax_form": "submitted a tax form", "has_vendor_approval": "has vendor approval", "has_risk_note": "has a risk note", "may_receive_purchase_order": "may receive a purchase order", "needs_procurement_review": "needs procurement review"}, "vocabulary_predicates": ["has_risk_note", "has_vendor_approval", "may_receive_purchase_order", "needs_procurement_review", "submitted_tax_form"], "query_atom": "has_risk_note::Vendor A", "target_statement": "Vendor A has a risk note.", "gold_reason_type": "explicit_negative_fact", "symbolic": {"positive_atoms": ["submitted_tax_form::Vendor C"], "negative_atoms": ["has_risk_note::Vendor A"], "rules": [], "query_atom": "has_risk_note::Vendor A", "closure_atoms": ["submitted_tax_form::Vendor C"]}, "source_reports": {"agent_a": {"positive": ["Vendor C submitted a tax form."], "negative": ["Vendor A does not has a risk note."]}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "explicit_negative", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_risk_note, has_vendor_approval, may_receive_purchase_order, needs_procurement_review, submitted_tax_form.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- has_risk_note: has a risk note\n- has_vendor_approval: has vendor approval\n- may_receive_purchase_order: may receive a purchase order\n- needs_procurement_review: needs procurement review\n- submitted_tax_form: submitted a tax form\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- Vendor C submitted a tax form.\nNegative facts:\n- Vendor A does not has a risk note.\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: has_risk_note::Vendor A.\nTarget statement: Vendor A has a risk note.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "procurement_review_0318_explicit_negative_open__lcwa__multi_agent", "source_id": "procurement_review_0318_explicit_negative_open__lcwa", "base_id": "procurement_review_0318_explicit_negative_open", "split": "dev", "domain": "procurement_review", "family": "explicit_negative_open", "subset": "control_explicit_negative", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_vendor_approval", "may_receive_purchase_order"], "agent_a_complete_predicates": ["has_vendor_approval", "may_receive_purchase_order"], "agent_b_open_predicates": ["has_risk_note", "needs_procurement_review", "submitted_tax_form"], "facts_positive": ["Vendor C submitted a tax form."], "facts_negative": ["Vendor A does not has a risk note."], "rules_natural": [], "predicate_glossary": {"submitted_tax_form": "submitted a tax form", "has_vendor_approval": "has vendor approval", "has_risk_note": "has a risk note", "may_receive_purchase_order": "may receive a purchase order", "needs_procurement_review": "needs procurement review"}, "vocabulary_predicates": ["has_risk_note", "has_vendor_approval", "may_receive_purchase_order", "needs_procurement_review", "submitted_tax_form"], "query_atom": "has_risk_note::Vendor A", "target_statement": "Vendor A has a risk note.", "gold_reason_type": "explicit_negative_fact", "symbolic": {"positive_atoms": ["submitted_tax_form::Vendor C"], "negative_atoms": ["has_risk_note::Vendor A"], "rules": [], "query_atom": "has_risk_note::Vendor A", "closure_atoms": ["submitted_tax_form::Vendor C"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Vendor C submitted a tax form."], "negative": ["Vendor A does not has a risk note."]}}, "gold_truth_value": "false", "gold_source_used": "agent_b", "gold_closure_handling": "explicit_negative", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_vendor_approval, may_receive_purchase_order.\nAgent B open/incomplete predicates: has_risk_note, needs_procurement_review, submitted_tax_form.\n\nPredicate glossary:\n- has_risk_note: has a risk note\n- has_vendor_approval: has vendor approval\n- may_receive_purchase_order: may receive a purchase order\n- needs_procurement_review: needs procurement review\n- submitted_tax_form: submitted a tax form\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Vendor C submitted a tax form.\nNegative facts:\n- Vendor A does not has a risk note.\n\nTarget atom: has_risk_note::Vendor A.\nTarget statement: Vendor A has a risk note.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "procurement_review_0318_explicit_negative_open__owa__multi_agent", "source_id": "procurement_review_0318_explicit_negative_open__owa", "base_id": "procurement_review_0318_explicit_negative_open", "split": "dev", "domain": "procurement_review", "family": "explicit_negative_open", "subset": "control_explicit_negative", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["has_risk_note", "has_vendor_approval", "may_receive_purchase_order", "needs_procurement_review", "submitted_tax_form"], "facts_positive": ["Vendor C submitted a tax form."], "facts_negative": ["Vendor A does not has a risk note."], "rules_natural": [], "predicate_glossary": {"submitted_tax_form": "submitted a tax form", "has_vendor_approval": "has vendor approval", "has_risk_note": "has a risk note", "may_receive_purchase_order": "may receive a purchase order", "needs_procurement_review": "needs procurement review"}, "vocabulary_predicates": ["has_risk_note", "has_vendor_approval", "may_receive_purchase_order", "needs_procurement_review", "submitted_tax_form"], "query_atom": "has_risk_note::Vendor A", "target_statement": "Vendor A has a risk note.", "gold_reason_type": "explicit_negative_fact", "symbolic": {"positive_atoms": ["submitted_tax_form::Vendor C"], "negative_atoms": ["has_risk_note::Vendor A"], "rules": [], "query_atom": "has_risk_note::Vendor A", "closure_atoms": ["submitted_tax_form::Vendor C"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Vendor C submitted a tax form."], "negative": ["Vendor A does not has a risk note."]}}, "gold_truth_value": "false", "gold_source_used": "agent_b", "gold_closure_handling": "explicit_negative", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: has_risk_note, has_vendor_approval, may_receive_purchase_order, needs_procurement_review, submitted_tax_form.\n\nPredicate glossary:\n- has_risk_note: has a risk note\n- has_vendor_approval: has vendor approval\n- may_receive_purchase_order: may receive a purchase order\n- needs_procurement_review: needs procurement review\n- submitted_tax_form: submitted a tax form\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Vendor C submitted a tax form.\nNegative facts:\n- Vendor A does not has a risk note.\n\nTarget atom: has_risk_note::Vendor A.\nTarget statement: Vendor A has a risk note.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "procurement_review_0319_explicit_negative_open__cwa__multi_agent", "source_id": "procurement_review_0319_explicit_negative_open__cwa", "base_id": "procurement_review_0319_explicit_negative_open", "split": "train", "domain": "procurement_review", "family": "explicit_negative_open", "subset": "control_explicit_negative", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["has_risk_note", "has_vendor_approval", "may_receive_purchase_order", "needs_procurement_review", "submitted_tax_form"], "agent_a_complete_predicates": ["has_risk_note", "has_vendor_approval", "may_receive_purchase_order", "needs_procurement_review", "submitted_tax_form"], "agent_b_open_predicates": [], "facts_positive": ["Vendor D submitted a tax form."], "facts_negative": ["Vendor B does not has a risk note."], "rules_natural": [], "predicate_glossary": {"submitted_tax_form": "submitted a tax form", "has_vendor_approval": "has vendor approval", "has_risk_note": "has a risk note", "may_receive_purchase_order": "may receive a purchase order", "needs_procurement_review": "needs procurement review"}, "vocabulary_predicates": ["has_risk_note", "has_vendor_approval", "may_receive_purchase_order", "needs_procurement_review", "submitted_tax_form"], "query_atom": "has_risk_note::Vendor B", "target_statement": "Vendor B has a risk note.", "gold_reason_type": "explicit_negative_fact", "symbolic": {"positive_atoms": ["submitted_tax_form::Vendor D"], "negative_atoms": ["has_risk_note::Vendor B"], "rules": [], "query_atom": "has_risk_note::Vendor B", "closure_atoms": ["submitted_tax_form::Vendor D"]}, "source_reports": {"agent_a": {"positive": ["Vendor D submitted a tax form."], "negative": ["Vendor B does not has a risk note."]}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "explicit_negative", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_risk_note, has_vendor_approval, may_receive_purchase_order, needs_procurement_review, submitted_tax_form.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- has_risk_note: has a risk note\n- has_vendor_approval: has vendor approval\n- may_receive_purchase_order: may receive a purchase order\n- needs_procurement_review: needs procurement review\n- submitted_tax_form: submitted a tax form\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- Vendor D submitted a tax form.\nNegative facts:\n- Vendor B does not has a risk note.\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: has_risk_note::Vendor B.\nTarget statement: Vendor B has a risk note.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "procurement_review_0319_explicit_negative_open__lcwa__multi_agent", "source_id": "procurement_review_0319_explicit_negative_open__lcwa", "base_id": "procurement_review_0319_explicit_negative_open", "split": "train", "domain": "procurement_review", "family": "explicit_negative_open", "subset": "control_explicit_negative", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_vendor_approval", "may_receive_purchase_order"], "agent_a_complete_predicates": ["has_vendor_approval", "may_receive_purchase_order"], "agent_b_open_predicates": ["has_risk_note", "needs_procurement_review", "submitted_tax_form"], "facts_positive": ["Vendor D submitted a tax form."], "facts_negative": ["Vendor B does not has a risk note."], "rules_natural": [], "predicate_glossary": {"submitted_tax_form": "submitted a tax form", "has_vendor_approval": "has vendor approval", "has_risk_note": "has a risk note", "may_receive_purchase_order": "may receive a purchase order", "needs_procurement_review": "needs procurement review"}, "vocabulary_predicates": ["has_risk_note", "has_vendor_approval", "may_receive_purchase_order", "needs_procurement_review", "submitted_tax_form"], "query_atom": "has_risk_note::Vendor B", "target_statement": "Vendor B has a risk note.", "gold_reason_type": "explicit_negative_fact", "symbolic": {"positive_atoms": ["submitted_tax_form::Vendor D"], "negative_atoms": ["has_risk_note::Vendor B"], "rules": [], "query_atom": "has_risk_note::Vendor B", "closure_atoms": ["submitted_tax_form::Vendor D"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Vendor D submitted a tax form."], "negative": ["Vendor B does not has a risk note."]}}, "gold_truth_value": "false", "gold_source_used": "agent_b", "gold_closure_handling": "explicit_negative", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_vendor_approval, may_receive_purchase_order.\nAgent B open/incomplete predicates: has_risk_note, needs_procurement_review, submitted_tax_form.\n\nPredicate glossary:\n- has_risk_note: has a risk note\n- has_vendor_approval: has vendor approval\n- may_receive_purchase_order: may receive a purchase order\n- needs_procurement_review: needs procurement review\n- submitted_tax_form: submitted a tax form\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Vendor D submitted a tax form.\nNegative facts:\n- Vendor B does not has a risk note.\n\nTarget atom: has_risk_note::Vendor B.\nTarget statement: Vendor B has a risk note.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "procurement_review_0319_explicit_negative_open__owa__multi_agent", "source_id": "procurement_review_0319_explicit_negative_open__owa", "base_id": "procurement_review_0319_explicit_negative_open", "split": "train", "domain": "procurement_review", "family": "explicit_negative_open", "subset": "control_explicit_negative", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["has_risk_note", "has_vendor_approval", "may_receive_purchase_order", "needs_procurement_review", "submitted_tax_form"], "facts_positive": ["Vendor D submitted a tax form."], "facts_negative": ["Vendor B does not has a risk note."], "rules_natural": [], "predicate_glossary": {"submitted_tax_form": "submitted a tax form", "has_vendor_approval": "has vendor approval", "has_risk_note": "has a risk note", "may_receive_purchase_order": "may receive a purchase order", "needs_procurement_review": "needs procurement review"}, "vocabulary_predicates": ["has_risk_note", "has_vendor_approval", "may_receive_purchase_order", "needs_procurement_review", "submitted_tax_form"], "query_atom": "has_risk_note::Vendor B", "target_statement": "Vendor B has a risk note.", "gold_reason_type": "explicit_negative_fact", "symbolic": {"positive_atoms": ["submitted_tax_form::Vendor D"], "negative_atoms": ["has_risk_note::Vendor B"], "rules": [], "query_atom": "has_risk_note::Vendor B", "closure_atoms": ["submitted_tax_form::Vendor D"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Vendor D submitted a tax form."], "negative": ["Vendor B does not has a risk note."]}}, "gold_truth_value": "false", "gold_source_used": "agent_b", "gold_closure_handling": "explicit_negative", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: has_risk_note, has_vendor_approval, may_receive_purchase_order, needs_procurement_review, submitted_tax_form.\n\nPredicate glossary:\n- has_risk_note: has a risk note\n- has_vendor_approval: has vendor approval\n- may_receive_purchase_order: may receive a purchase order\n- needs_procurement_review: needs procurement review\n- submitted_tax_form: submitted a tax form\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Vendor D submitted a tax form.\nNegative facts:\n- Vendor B does not has a risk note.\n\nTarget atom: has_risk_note::Vendor B.\nTarget statement: Vendor B has a risk note.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "robotics_operations_0160_closed_missing_direct__cwa__multi_agent", "source_id": "robotics_operations_0160_closed_missing_direct__cwa", "base_id": "robotics_operations_0160_closed_missing_direct", "split": "train", "domain": "robotics_operations", "family": "closed_missing_direct", "subset": "core_contrastive", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["has_field_note", "has_shift_clearance", "may_start_shift", "needs_operator_review", "passed_diagnostics"], "agent_a_complete_predicates": ["has_field_note", "has_shift_clearance", "may_start_shift", "needs_operator_review", "passed_diagnostics"], "agent_b_open_predicates": [], "facts_positive": ["Rover 5 passed diagnostics."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_diagnostics": "passed diagnostics", "has_shift_clearance": "has shift clearance", "has_field_note": "has a field note", "may_start_shift": "may start its shift", "needs_operator_review": "needs operator review"}, "vocabulary_predicates": ["has_field_note", "has_shift_clearance", "may_start_shift", "needs_operator_review", "passed_diagnostics"], "query_atom": "has_shift_clearance::Rover 5", "target_statement": "Rover 5 has shift clearance.", "gold_reason_type": "global_closed_world_underivable", "symbolic": {"positive_atoms": ["passed_diagnostics::Rover 5"], "negative_atoms": [], "rules": [], "query_atom": "has_shift_clearance::Rover 5", "closure_atoms": ["passed_diagnostics::Rover 5"]}, "source_reports": {"agent_a": {"positive": ["Rover 5 passed diagnostics."], "negative": []}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "closed_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_field_note, has_shift_clearance, may_start_shift, needs_operator_review, passed_diagnostics.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- has_field_note: has a field note\n- has_shift_clearance: has shift clearance\n- may_start_shift: may start its shift\n- needs_operator_review: needs operator review\n- passed_diagnostics: passed diagnostics\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- Rover 5 passed diagnostics.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: has_shift_clearance::Rover 5.\nTarget statement: Rover 5 has shift clearance.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "robotics_operations_0160_closed_missing_direct__lcwa__multi_agent", "source_id": "robotics_operations_0160_closed_missing_direct__lcwa", "base_id": "robotics_operations_0160_closed_missing_direct", "split": "train", "domain": "robotics_operations", "family": "closed_missing_direct", "subset": "core_contrastive", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_shift_clearance"], "agent_a_complete_predicates": ["has_shift_clearance"], "agent_b_open_predicates": ["has_field_note", "may_start_shift", "needs_operator_review", "passed_diagnostics"], "facts_positive": ["Rover 5 passed diagnostics."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_diagnostics": "passed diagnostics", "has_shift_clearance": "has shift clearance", "has_field_note": "has a field note", "may_start_shift": "may start its shift", "needs_operator_review": "needs operator review"}, "vocabulary_predicates": ["has_field_note", "has_shift_clearance", "may_start_shift", "needs_operator_review", "passed_diagnostics"], "query_atom": "has_shift_clearance::Rover 5", "target_statement": "Rover 5 has shift clearance.", "gold_reason_type": "local_closed_world_underivable", "symbolic": {"positive_atoms": ["passed_diagnostics::Rover 5"], "negative_atoms": [], "rules": [], "query_atom": "has_shift_clearance::Rover 5", "closure_atoms": ["passed_diagnostics::Rover 5"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Rover 5 passed diagnostics."], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "closed_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_shift_clearance.\nAgent B open/incomplete predicates: has_field_note, may_start_shift, needs_operator_review, passed_diagnostics.\n\nPredicate glossary:\n- has_field_note: has a field note\n- has_shift_clearance: has shift clearance\n- may_start_shift: may start its shift\n- needs_operator_review: needs operator review\n- passed_diagnostics: passed diagnostics\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Rover 5 passed diagnostics.\nNegative facts:\n- none\n\nTarget atom: has_shift_clearance::Rover 5.\nTarget statement: Rover 5 has shift clearance.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "robotics_operations_0160_closed_missing_direct__owa__multi_agent", "source_id": "robotics_operations_0160_closed_missing_direct__owa", "base_id": "robotics_operations_0160_closed_missing_direct", "split": "train", "domain": "robotics_operations", "family": "closed_missing_direct", "subset": "core_contrastive", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["has_field_note", "has_shift_clearance", "may_start_shift", "needs_operator_review", "passed_diagnostics"], "facts_positive": ["Rover 5 passed diagnostics."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_diagnostics": "passed diagnostics", "has_shift_clearance": "has shift clearance", "has_field_note": "has a field note", "may_start_shift": "may start its shift", "needs_operator_review": "needs operator review"}, "vocabulary_predicates": ["has_field_note", "has_shift_clearance", "may_start_shift", "needs_operator_review", "passed_diagnostics"], "query_atom": "has_shift_clearance::Rover 5", "target_statement": "Rover 5 has shift clearance.", "gold_reason_type": "open_world_underivable", "symbolic": {"positive_atoms": ["passed_diagnostics::Rover 5"], "negative_atoms": [], "rules": [], "query_atom": "has_shift_clearance::Rover 5", "closure_atoms": ["passed_diagnostics::Rover 5"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Rover 5 passed diagnostics."], "negative": []}}, "gold_truth_value": "unknown", "gold_source_used": "agent_b", "gold_closure_handling": "open_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: has_field_note, has_shift_clearance, may_start_shift, needs_operator_review, passed_diagnostics.\n\nPredicate glossary:\n- has_field_note: has a field note\n- has_shift_clearance: has shift clearance\n- may_start_shift: may start its shift\n- needs_operator_review: needs operator review\n- passed_diagnostics: passed diagnostics\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Rover 5 passed diagnostics.\nNegative facts:\n- none\n\nTarget atom: has_shift_clearance::Rover 5.\nTarget statement: Rover 5 has shift clearance.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "robotics_operations_0167_open_missing_direct__cwa__multi_agent", "source_id": "robotics_operations_0167_open_missing_direct__cwa", "base_id": "robotics_operations_0167_open_missing_direct", "split": "train", "domain": "robotics_operations", "family": "open_missing_direct", "subset": "core_contrastive", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["has_field_note", "has_shift_clearance", "may_start_shift", "needs_operator_review", "passed_diagnostics"], "agent_a_complete_predicates": ["has_field_note", "has_shift_clearance", "may_start_shift", "needs_operator_review", "passed_diagnostics"], "agent_b_open_predicates": [], "facts_positive": ["Rover 6 passed diagnostics."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_diagnostics": "passed diagnostics", "has_shift_clearance": "has shift clearance", "has_field_note": "has a field note", "may_start_shift": "may start its shift", "needs_operator_review": "needs operator review"}, "vocabulary_predicates": ["has_field_note", "has_shift_clearance", "may_start_shift", "needs_operator_review", "passed_diagnostics"], "query_atom": "has_field_note::Rover 6", "target_statement": "Rover 6 has a field note.", "gold_reason_type": "global_closed_world_underivable", "symbolic": {"positive_atoms": ["passed_diagnostics::Rover 6"], "negative_atoms": [], "rules": [], "query_atom": "has_field_note::Rover 6", "closure_atoms": ["passed_diagnostics::Rover 6"]}, "source_reports": {"agent_a": {"positive": ["Rover 6 passed diagnostics."], "negative": []}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "closed_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_field_note, has_shift_clearance, may_start_shift, needs_operator_review, passed_diagnostics.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- has_field_note: has a field note\n- has_shift_clearance: has shift clearance\n- may_start_shift: may start its shift\n- needs_operator_review: needs operator review\n- passed_diagnostics: passed diagnostics\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- Rover 6 passed diagnostics.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: has_field_note::Rover 6.\nTarget statement: Rover 6 has a field note.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "robotics_operations_0167_open_missing_direct__lcwa__multi_agent", "source_id": "robotics_operations_0167_open_missing_direct__lcwa", "base_id": "robotics_operations_0167_open_missing_direct", "split": "train", "domain": "robotics_operations", "family": "open_missing_direct", "subset": "core_contrastive", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_shift_clearance", "may_start_shift"], "agent_a_complete_predicates": ["has_shift_clearance", "may_start_shift"], "agent_b_open_predicates": ["has_field_note", "needs_operator_review", "passed_diagnostics"], "facts_positive": ["Rover 6 passed diagnostics."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_diagnostics": "passed diagnostics", "has_shift_clearance": "has shift clearance", "has_field_note": "has a field note", "may_start_shift": "may start its shift", "needs_operator_review": "needs operator review"}, "vocabulary_predicates": ["has_field_note", "has_shift_clearance", "may_start_shift", "needs_operator_review", "passed_diagnostics"], "query_atom": "has_field_note::Rover 6", "target_statement": "Rover 6 has a field note.", "gold_reason_type": "local_open_world_underivable", "symbolic": {"positive_atoms": ["passed_diagnostics::Rover 6"], "negative_atoms": [], "rules": [], "query_atom": "has_field_note::Rover 6", "closure_atoms": ["passed_diagnostics::Rover 6"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Rover 6 passed diagnostics."], "negative": []}}, "gold_truth_value": "unknown", "gold_source_used": "agent_b", "gold_closure_handling": "open_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_shift_clearance, may_start_shift.\nAgent B open/incomplete predicates: has_field_note, needs_operator_review, passed_diagnostics.\n\nPredicate glossary:\n- has_field_note: has a field note\n- has_shift_clearance: has shift clearance\n- may_start_shift: may start its shift\n- needs_operator_review: needs operator review\n- passed_diagnostics: passed diagnostics\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Rover 6 passed diagnostics.\nNegative facts:\n- none\n\nTarget atom: has_field_note::Rover 6.\nTarget statement: Rover 6 has a field note.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "robotics_operations_0167_open_missing_direct__owa__multi_agent", "source_id": "robotics_operations_0167_open_missing_direct__owa", "base_id": "robotics_operations_0167_open_missing_direct", "split": "train", "domain": "robotics_operations", "family": "open_missing_direct", "subset": "core_contrastive", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["has_field_note", "has_shift_clearance", "may_start_shift", "needs_operator_review", "passed_diagnostics"], "facts_positive": ["Rover 6 passed diagnostics."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_diagnostics": "passed diagnostics", "has_shift_clearance": "has shift clearance", "has_field_note": "has a field note", "may_start_shift": "may start its shift", "needs_operator_review": "needs operator review"}, "vocabulary_predicates": ["has_field_note", "has_shift_clearance", "may_start_shift", "needs_operator_review", "passed_diagnostics"], "query_atom": "has_field_note::Rover 6", "target_statement": "Rover 6 has a field note.", "gold_reason_type": "open_world_underivable", "symbolic": {"positive_atoms": ["passed_diagnostics::Rover 6"], "negative_atoms": [], "rules": [], "query_atom": "has_field_note::Rover 6", "closure_atoms": ["passed_diagnostics::Rover 6"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Rover 6 passed diagnostics."], "negative": []}}, "gold_truth_value": "unknown", "gold_source_used": "agent_b", "gold_closure_handling": "open_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: has_field_note, has_shift_clearance, may_start_shift, needs_operator_review, passed_diagnostics.\n\nPredicate glossary:\n- has_field_note: has a field note\n- has_shift_clearance: has shift clearance\n- may_start_shift: may start its shift\n- needs_operator_review: needs operator review\n- passed_diagnostics: passed diagnostics\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Rover 6 passed diagnostics.\nNegative facts:\n- none\n\nTarget atom: has_field_note::Rover 6.\nTarget statement: Rover 6 has a field note.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "robotics_operations_0168_closed_missing_with_open_distractor__cwa__multi_agent", "source_id": "robotics_operations_0168_closed_missing_with_open_distractor__cwa", "base_id": "robotics_operations_0168_closed_missing_with_open_distractor", "split": "train", "domain": "robotics_operations", "family": "closed_missing_with_open_distractor", "subset": "core_contrastive", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["has_field_note", "has_shift_clearance", "may_start_shift", "needs_operator_review", "passed_diagnostics"], "agent_a_complete_predicates": ["has_field_note", "has_shift_clearance", "may_start_shift", "needs_operator_review", "passed_diagnostics"], "agent_b_open_predicates": [], "facts_positive": ["Rover 2 has a field note.", "Rover 1 passed diagnostics."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_diagnostics": "passed diagnostics", "has_shift_clearance": "has shift clearance", "has_field_note": "has a field note", "may_start_shift": "may start its shift", "needs_operator_review": "needs operator review"}, "vocabulary_predicates": ["has_field_note", "has_shift_clearance", "may_start_shift", "needs_operator_review", "passed_diagnostics"], "query_atom": "has_shift_clearance::Rover 1", "target_statement": "Rover 1 has shift clearance.", "gold_reason_type": "global_closed_world_underivable", "symbolic": {"positive_atoms": ["has_field_note::Rover 2", "passed_diagnostics::Rover 1"], "negative_atoms": [], "rules": [], "query_atom": "has_shift_clearance::Rover 1", "closure_atoms": ["has_field_note::Rover 2", "passed_diagnostics::Rover 1"]}, "source_reports": {"agent_a": {"positive": ["Rover 2 has a field note.", "Rover 1 passed diagnostics."], "negative": []}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "closed_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_field_note, has_shift_clearance, may_start_shift, needs_operator_review, passed_diagnostics.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- has_field_note: has a field note\n- has_shift_clearance: has shift clearance\n- may_start_shift: may start its shift\n- needs_operator_review: needs operator review\n- passed_diagnostics: passed diagnostics\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- Rover 2 has a field note.\n- Rover 1 passed diagnostics.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: has_shift_clearance::Rover 1.\nTarget statement: Rover 1 has shift clearance.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "robotics_operations_0168_closed_missing_with_open_distractor__lcwa__multi_agent", "source_id": "robotics_operations_0168_closed_missing_with_open_distractor__lcwa", "base_id": "robotics_operations_0168_closed_missing_with_open_distractor", "split": "train", "domain": "robotics_operations", "family": "closed_missing_with_open_distractor", "subset": "core_contrastive", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_shift_clearance"], "agent_a_complete_predicates": ["has_shift_clearance"], "agent_b_open_predicates": ["has_field_note", "may_start_shift", "needs_operator_review", "passed_diagnostics"], "facts_positive": ["Rover 2 has a field note.", "Rover 1 passed diagnostics."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_diagnostics": "passed diagnostics", "has_shift_clearance": "has shift clearance", "has_field_note": "has a field note", "may_start_shift": "may start its shift", "needs_operator_review": "needs operator review"}, "vocabulary_predicates": ["has_field_note", "has_shift_clearance", "may_start_shift", "needs_operator_review", "passed_diagnostics"], "query_atom": "has_shift_clearance::Rover 1", "target_statement": "Rover 1 has shift clearance.", "gold_reason_type": "local_closed_world_underivable", "symbolic": {"positive_atoms": ["has_field_note::Rover 2", "passed_diagnostics::Rover 1"], "negative_atoms": [], "rules": [], "query_atom": "has_shift_clearance::Rover 1", "closure_atoms": ["has_field_note::Rover 2", "passed_diagnostics::Rover 1"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Rover 2 has a field note.", "Rover 1 passed diagnostics."], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "closed_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_shift_clearance.\nAgent B open/incomplete predicates: has_field_note, may_start_shift, needs_operator_review, passed_diagnostics.\n\nPredicate glossary:\n- has_field_note: has a field note\n- has_shift_clearance: has shift clearance\n- may_start_shift: may start its shift\n- needs_operator_review: needs operator review\n- passed_diagnostics: passed diagnostics\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Rover 2 has a field note.\n- Rover 1 passed diagnostics.\nNegative facts:\n- none\n\nTarget atom: has_shift_clearance::Rover 1.\nTarget statement: Rover 1 has shift clearance.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "robotics_operations_0168_closed_missing_with_open_distractor__owa__multi_agent", "source_id": "robotics_operations_0168_closed_missing_with_open_distractor__owa", "base_id": "robotics_operations_0168_closed_missing_with_open_distractor", "split": "train", "domain": "robotics_operations", "family": "closed_missing_with_open_distractor", "subset": "core_contrastive", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["has_field_note", "has_shift_clearance", "may_start_shift", "needs_operator_review", "passed_diagnostics"], "facts_positive": ["Rover 2 has a field note.", "Rover 1 passed diagnostics."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_diagnostics": "passed diagnostics", "has_shift_clearance": "has shift clearance", "has_field_note": "has a field note", "may_start_shift": "may start its shift", "needs_operator_review": "needs operator review"}, "vocabulary_predicates": ["has_field_note", "has_shift_clearance", "may_start_shift", "needs_operator_review", "passed_diagnostics"], "query_atom": "has_shift_clearance::Rover 1", "target_statement": "Rover 1 has shift clearance.", "gold_reason_type": "open_world_underivable", "symbolic": {"positive_atoms": ["has_field_note::Rover 2", "passed_diagnostics::Rover 1"], "negative_atoms": [], "rules": [], "query_atom": "has_shift_clearance::Rover 1", "closure_atoms": ["has_field_note::Rover 2", "passed_diagnostics::Rover 1"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Rover 2 has a field note.", "Rover 1 passed diagnostics."], "negative": []}}, "gold_truth_value": "unknown", "gold_source_used": "agent_b", "gold_closure_handling": "open_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: has_field_note, has_shift_clearance, may_start_shift, needs_operator_review, passed_diagnostics.\n\nPredicate glossary:\n- has_field_note: has a field note\n- has_shift_clearance: has shift clearance\n- may_start_shift: may start its shift\n- needs_operator_review: needs operator review\n- passed_diagnostics: passed diagnostics\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Rover 2 has a field note.\n- Rover 1 passed diagnostics.\nNegative facts:\n- none\n\nTarget atom: has_shift_clearance::Rover 1.\nTarget statement: Rover 1 has shift clearance.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "robotics_operations_0170_closed_missing_with_open_distractor__cwa__multi_agent", "source_id": "robotics_operations_0170_closed_missing_with_open_distractor__cwa", "base_id": "robotics_operations_0170_closed_missing_with_open_distractor", "split": "train", "domain": "robotics_operations", "family": "closed_missing_with_open_distractor", "subset": "core_contrastive", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["has_field_note", "has_shift_clearance", "may_start_shift", "needs_operator_review", "passed_diagnostics"], "agent_a_complete_predicates": ["has_field_note", "has_shift_clearance", "may_start_shift", "needs_operator_review", "passed_diagnostics"], "agent_b_open_predicates": [], "facts_positive": ["Rover 4 has a field note.", "Rover 3 passed diagnostics."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_diagnostics": "passed diagnostics", "has_shift_clearance": "has shift clearance", "has_field_note": "has a field note", "may_start_shift": "may start its shift", "needs_operator_review": "needs operator review"}, "vocabulary_predicates": ["has_field_note", "has_shift_clearance", "may_start_shift", "needs_operator_review", "passed_diagnostics"], "query_atom": "has_shift_clearance::Rover 3", "target_statement": "Rover 3 has shift clearance.", "gold_reason_type": "global_closed_world_underivable", "symbolic": {"positive_atoms": ["has_field_note::Rover 4", "passed_diagnostics::Rover 3"], "negative_atoms": [], "rules": [], "query_atom": "has_shift_clearance::Rover 3", "closure_atoms": ["has_field_note::Rover 4", "passed_diagnostics::Rover 3"]}, "source_reports": {"agent_a": {"positive": ["Rover 4 has a field note.", "Rover 3 passed diagnostics."], "negative": []}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "closed_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_field_note, has_shift_clearance, may_start_shift, needs_operator_review, passed_diagnostics.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- has_field_note: has a field note\n- has_shift_clearance: has shift clearance\n- may_start_shift: may start its shift\n- needs_operator_review: needs operator review\n- passed_diagnostics: passed diagnostics\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- Rover 4 has a field note.\n- Rover 3 passed diagnostics.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: has_shift_clearance::Rover 3.\nTarget statement: Rover 3 has shift clearance.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "robotics_operations_0170_closed_missing_with_open_distractor__lcwa__multi_agent", "source_id": "robotics_operations_0170_closed_missing_with_open_distractor__lcwa", "base_id": "robotics_operations_0170_closed_missing_with_open_distractor", "split": "train", "domain": "robotics_operations", "family": "closed_missing_with_open_distractor", "subset": "core_contrastive", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_shift_clearance"], "agent_a_complete_predicates": ["has_shift_clearance"], "agent_b_open_predicates": ["has_field_note", "may_start_shift", "needs_operator_review", "passed_diagnostics"], "facts_positive": ["Rover 4 has a field note.", "Rover 3 passed diagnostics."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_diagnostics": "passed diagnostics", "has_shift_clearance": "has shift clearance", "has_field_note": "has a field note", "may_start_shift": "may start its shift", "needs_operator_review": "needs operator review"}, "vocabulary_predicates": ["has_field_note", "has_shift_clearance", "may_start_shift", "needs_operator_review", "passed_diagnostics"], "query_atom": "has_shift_clearance::Rover 3", "target_statement": "Rover 3 has shift clearance.", "gold_reason_type": "local_closed_world_underivable", "symbolic": {"positive_atoms": ["has_field_note::Rover 4", "passed_diagnostics::Rover 3"], "negative_atoms": [], "rules": [], "query_atom": "has_shift_clearance::Rover 3", "closure_atoms": ["has_field_note::Rover 4", "passed_diagnostics::Rover 3"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Rover 4 has a field note.", "Rover 3 passed diagnostics."], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "closed_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_shift_clearance.\nAgent B open/incomplete predicates: has_field_note, may_start_shift, needs_operator_review, passed_diagnostics.\n\nPredicate glossary:\n- has_field_note: has a field note\n- has_shift_clearance: has shift clearance\n- may_start_shift: may start its shift\n- needs_operator_review: needs operator review\n- passed_diagnostics: passed diagnostics\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Rover 4 has a field note.\n- Rover 3 passed diagnostics.\nNegative facts:\n- none\n\nTarget atom: has_shift_clearance::Rover 3.\nTarget statement: Rover 3 has shift clearance.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "robotics_operations_0170_closed_missing_with_open_distractor__owa__multi_agent", "source_id": "robotics_operations_0170_closed_missing_with_open_distractor__owa", "base_id": "robotics_operations_0170_closed_missing_with_open_distractor", "split": "train", "domain": "robotics_operations", "family": "closed_missing_with_open_distractor", "subset": "core_contrastive", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["has_field_note", "has_shift_clearance", "may_start_shift", "needs_operator_review", "passed_diagnostics"], "facts_positive": ["Rover 4 has a field note.", "Rover 3 passed diagnostics."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_diagnostics": "passed diagnostics", "has_shift_clearance": "has shift clearance", "has_field_note": "has a field note", "may_start_shift": "may start its shift", "needs_operator_review": "needs operator review"}, "vocabulary_predicates": ["has_field_note", "has_shift_clearance", "may_start_shift", "needs_operator_review", "passed_diagnostics"], "query_atom": "has_shift_clearance::Rover 3", "target_statement": "Rover 3 has shift clearance.", "gold_reason_type": "open_world_underivable", "symbolic": {"positive_atoms": ["has_field_note::Rover 4", "passed_diagnostics::Rover 3"], "negative_atoms": [], "rules": [], "query_atom": "has_shift_clearance::Rover 3", "closure_atoms": ["has_field_note::Rover 4", "passed_diagnostics::Rover 3"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Rover 4 has a field note.", "Rover 3 passed diagnostics."], "negative": []}}, "gold_truth_value": "unknown", "gold_source_used": "agent_b", "gold_closure_handling": "open_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: has_field_note, has_shift_clearance, may_start_shift, needs_operator_review, passed_diagnostics.\n\nPredicate glossary:\n- has_field_note: has a field note\n- has_shift_clearance: has shift clearance\n- may_start_shift: may start its shift\n- needs_operator_review: needs operator review\n- passed_diagnostics: passed diagnostics\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Rover 4 has a field note.\n- Rover 3 passed diagnostics.\nNegative facts:\n- none\n\nTarget atom: has_shift_clearance::Rover 3.\nTarget statement: Rover 3 has shift clearance.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "robotics_operations_0172_open_missing_with_closed_distractor__cwa__multi_agent", "source_id": "robotics_operations_0172_open_missing_with_closed_distractor__cwa", "base_id": "robotics_operations_0172_open_missing_with_closed_distractor", "split": "train", "domain": "robotics_operations", "family": "open_missing_with_closed_distractor", "subset": "core_contrastive", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["has_field_note", "has_shift_clearance", "may_start_shift", "needs_operator_review", "passed_diagnostics"], "agent_a_complete_predicates": ["has_field_note", "has_shift_clearance", "may_start_shift", "needs_operator_review", "passed_diagnostics"], "agent_b_open_predicates": [], "facts_positive": ["Rover 6 has shift clearance.", "Rover 6 passed diagnostics."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_diagnostics": "passed diagnostics", "has_shift_clearance": "has shift clearance", "has_field_note": "has a field note", "may_start_shift": "may start its shift", "needs_operator_review": "needs operator review"}, "vocabulary_predicates": ["has_field_note", "has_shift_clearance", "may_start_shift", "needs_operator_review", "passed_diagnostics"], "query_atom": "has_field_note::Rover 5", "target_statement": "Rover 5 has a field note.", "gold_reason_type": "global_closed_world_underivable", "symbolic": {"positive_atoms": ["has_shift_clearance::Rover 6", "passed_diagnostics::Rover 6"], "negative_atoms": [], "rules": [], "query_atom": "has_field_note::Rover 5", "closure_atoms": ["has_shift_clearance::Rover 6", "passed_diagnostics::Rover 6"]}, "source_reports": {"agent_a": {"positive": ["Rover 6 has shift clearance.", "Rover 6 passed diagnostics."], "negative": []}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "closed_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_field_note, has_shift_clearance, may_start_shift, needs_operator_review, passed_diagnostics.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- has_field_note: has a field note\n- has_shift_clearance: has shift clearance\n- may_start_shift: may start its shift\n- needs_operator_review: needs operator review\n- passed_diagnostics: passed diagnostics\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- Rover 6 has shift clearance.\n- Rover 6 passed diagnostics.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: has_field_note::Rover 5.\nTarget statement: Rover 5 has a field note.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "robotics_operations_0172_open_missing_with_closed_distractor__lcwa__multi_agent", "source_id": "robotics_operations_0172_open_missing_with_closed_distractor__lcwa", "base_id": "robotics_operations_0172_open_missing_with_closed_distractor", "split": "train", "domain": "robotics_operations", "family": "open_missing_with_closed_distractor", "subset": "core_contrastive", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_shift_clearance", "may_start_shift"], "agent_a_complete_predicates": ["has_shift_clearance", "may_start_shift"], "agent_b_open_predicates": ["has_field_note", "needs_operator_review", "passed_diagnostics"], "facts_positive": ["Rover 6 has shift clearance.", "Rover 6 passed diagnostics."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_diagnostics": "passed diagnostics", "has_shift_clearance": "has shift clearance", "has_field_note": "has a field note", "may_start_shift": "may start its shift", "needs_operator_review": "needs operator review"}, "vocabulary_predicates": ["has_field_note", "has_shift_clearance", "may_start_shift", "needs_operator_review", "passed_diagnostics"], "query_atom": "has_field_note::Rover 5", "target_statement": "Rover 5 has a field note.", "gold_reason_type": "local_open_world_underivable", "symbolic": {"positive_atoms": ["has_shift_clearance::Rover 6", "passed_diagnostics::Rover 6"], "negative_atoms": [], "rules": [], "query_atom": "has_field_note::Rover 5", "closure_atoms": ["has_shift_clearance::Rover 6", "passed_diagnostics::Rover 6"]}, "source_reports": {"agent_a": {"positive": ["Rover 6 has shift clearance."], "negative": []}, "agent_b": {"positive": ["Rover 6 passed diagnostics."], "negative": []}}, "gold_truth_value": "unknown", "gold_source_used": "agent_b", "gold_closure_handling": "open_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_shift_clearance, may_start_shift.\nAgent B open/incomplete predicates: has_field_note, needs_operator_review, passed_diagnostics.\n\nPredicate glossary:\n- has_field_note: has a field note\n- has_shift_clearance: has shift clearance\n- may_start_shift: may start its shift\n- needs_operator_review: needs operator review\n- passed_diagnostics: passed diagnostics\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- Rover 6 has shift clearance.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Rover 6 passed diagnostics.\nNegative facts:\n- none\n\nTarget atom: has_field_note::Rover 5.\nTarget statement: Rover 5 has a field note.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "robotics_operations_0172_open_missing_with_closed_distractor__owa__multi_agent", "source_id": "robotics_operations_0172_open_missing_with_closed_distractor__owa", "base_id": "robotics_operations_0172_open_missing_with_closed_distractor", "split": "train", "domain": "robotics_operations", "family": "open_missing_with_closed_distractor", "subset": "core_contrastive", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["has_field_note", "has_shift_clearance", "may_start_shift", "needs_operator_review", "passed_diagnostics"], "facts_positive": ["Rover 6 has shift clearance.", "Rover 6 passed diagnostics."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_diagnostics": "passed diagnostics", "has_shift_clearance": "has shift clearance", "has_field_note": "has a field note", "may_start_shift": "may start its shift", "needs_operator_review": "needs operator review"}, "vocabulary_predicates": ["has_field_note", "has_shift_clearance", "may_start_shift", "needs_operator_review", "passed_diagnostics"], "query_atom": "has_field_note::Rover 5", "target_statement": "Rover 5 has a field note.", "gold_reason_type": "open_world_underivable", "symbolic": {"positive_atoms": ["has_shift_clearance::Rover 6", "passed_diagnostics::Rover 6"], "negative_atoms": [], "rules": [], "query_atom": "has_field_note::Rover 5", "closure_atoms": ["has_shift_clearance::Rover 6", "passed_diagnostics::Rover 6"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Rover 6 has shift clearance.", "Rover 6 passed diagnostics."], "negative": []}}, "gold_truth_value": "unknown", "gold_source_used": "agent_b", "gold_closure_handling": "open_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: has_field_note, has_shift_clearance, may_start_shift, needs_operator_review, passed_diagnostics.\n\nPredicate glossary:\n- has_field_note: has a field note\n- has_shift_clearance: has shift clearance\n- may_start_shift: may start its shift\n- needs_operator_review: needs operator review\n- passed_diagnostics: passed diagnostics\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Rover 6 has shift clearance.\n- Rover 6 passed diagnostics.\nNegative facts:\n- none\n\nTarget atom: has_field_note::Rover 5.\nTarget statement: Rover 5 has a field note.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "robotics_operations_0173_open_missing_with_closed_distractor__cwa__multi_agent", "source_id": "robotics_operations_0173_open_missing_with_closed_distractor__cwa", "base_id": "robotics_operations_0173_open_missing_with_closed_distractor", "split": "test", "domain": "robotics_operations", "family": "open_missing_with_closed_distractor", "subset": "core_contrastive", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["has_field_note", "has_shift_clearance", "may_start_shift", "needs_operator_review", "passed_diagnostics"], "agent_a_complete_predicates": ["has_field_note", "has_shift_clearance", "may_start_shift", "needs_operator_review", "passed_diagnostics"], "agent_b_open_predicates": [], "facts_positive": ["Rover 1 has shift clearance.", "Rover 1 passed diagnostics."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_diagnostics": "passed diagnostics", "has_shift_clearance": "has shift clearance", "has_field_note": "has a field note", "may_start_shift": "may start its shift", "needs_operator_review": "needs operator review"}, "vocabulary_predicates": ["has_field_note", "has_shift_clearance", "may_start_shift", "needs_operator_review", "passed_diagnostics"], "query_atom": "has_field_note::Rover 6", "target_statement": "Rover 6 has a field note.", "gold_reason_type": "global_closed_world_underivable", "symbolic": {"positive_atoms": ["has_shift_clearance::Rover 1", "passed_diagnostics::Rover 1"], "negative_atoms": [], "rules": [], "query_atom": "has_field_note::Rover 6", "closure_atoms": ["has_shift_clearance::Rover 1", "passed_diagnostics::Rover 1"]}, "source_reports": {"agent_a": {"positive": ["Rover 1 has shift clearance.", "Rover 1 passed diagnostics."], "negative": []}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "closed_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_field_note, has_shift_clearance, may_start_shift, needs_operator_review, passed_diagnostics.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- has_field_note: has a field note\n- has_shift_clearance: has shift clearance\n- may_start_shift: may start its shift\n- needs_operator_review: needs operator review\n- passed_diagnostics: passed diagnostics\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- Rover 1 has shift clearance.\n- Rover 1 passed diagnostics.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: has_field_note::Rover 6.\nTarget statement: Rover 6 has a field note.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "robotics_operations_0173_open_missing_with_closed_distractor__lcwa__multi_agent", "source_id": "robotics_operations_0173_open_missing_with_closed_distractor__lcwa", "base_id": "robotics_operations_0173_open_missing_with_closed_distractor", "split": "test", "domain": "robotics_operations", "family": "open_missing_with_closed_distractor", "subset": "core_contrastive", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_shift_clearance", "may_start_shift"], "agent_a_complete_predicates": ["has_shift_clearance", "may_start_shift"], "agent_b_open_predicates": ["has_field_note", "needs_operator_review", "passed_diagnostics"], "facts_positive": ["Rover 1 has shift clearance.", "Rover 1 passed diagnostics."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_diagnostics": "passed diagnostics", "has_shift_clearance": "has shift clearance", "has_field_note": "has a field note", "may_start_shift": "may start its shift", "needs_operator_review": "needs operator review"}, "vocabulary_predicates": ["has_field_note", "has_shift_clearance", "may_start_shift", "needs_operator_review", "passed_diagnostics"], "query_atom": "has_field_note::Rover 6", "target_statement": "Rover 6 has a field note.", "gold_reason_type": "local_open_world_underivable", "symbolic": {"positive_atoms": ["has_shift_clearance::Rover 1", "passed_diagnostics::Rover 1"], "negative_atoms": [], "rules": [], "query_atom": "has_field_note::Rover 6", "closure_atoms": ["has_shift_clearance::Rover 1", "passed_diagnostics::Rover 1"]}, "source_reports": {"agent_a": {"positive": ["Rover 1 has shift clearance."], "negative": []}, "agent_b": {"positive": ["Rover 1 passed diagnostics."], "negative": []}}, "gold_truth_value": "unknown", "gold_source_used": "agent_b", "gold_closure_handling": "open_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_shift_clearance, may_start_shift.\nAgent B open/incomplete predicates: has_field_note, needs_operator_review, passed_diagnostics.\n\nPredicate glossary:\n- has_field_note: has a field note\n- has_shift_clearance: has shift clearance\n- may_start_shift: may start its shift\n- needs_operator_review: needs operator review\n- passed_diagnostics: passed diagnostics\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- Rover 1 has shift clearance.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Rover 1 passed diagnostics.\nNegative facts:\n- none\n\nTarget atom: has_field_note::Rover 6.\nTarget statement: Rover 6 has a field note.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "robotics_operations_0173_open_missing_with_closed_distractor__owa__multi_agent", "source_id": "robotics_operations_0173_open_missing_with_closed_distractor__owa", "base_id": "robotics_operations_0173_open_missing_with_closed_distractor", "split": "test", "domain": "robotics_operations", "family": "open_missing_with_closed_distractor", "subset": "core_contrastive", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["has_field_note", "has_shift_clearance", "may_start_shift", "needs_operator_review", "passed_diagnostics"], "facts_positive": ["Rover 1 has shift clearance.", "Rover 1 passed diagnostics."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_diagnostics": "passed diagnostics", "has_shift_clearance": "has shift clearance", "has_field_note": "has a field note", "may_start_shift": "may start its shift", "needs_operator_review": "needs operator review"}, "vocabulary_predicates": ["has_field_note", "has_shift_clearance", "may_start_shift", "needs_operator_review", "passed_diagnostics"], "query_atom": "has_field_note::Rover 6", "target_statement": "Rover 6 has a field note.", "gold_reason_type": "open_world_underivable", "symbolic": {"positive_atoms": ["has_shift_clearance::Rover 1", "passed_diagnostics::Rover 1"], "negative_atoms": [], "rules": [], "query_atom": "has_field_note::Rover 6", "closure_atoms": ["has_shift_clearance::Rover 1", "passed_diagnostics::Rover 1"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Rover 1 has shift clearance.", "Rover 1 passed diagnostics."], "negative": []}}, "gold_truth_value": "unknown", "gold_source_used": "agent_b", "gold_closure_handling": "open_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: has_field_note, has_shift_clearance, may_start_shift, needs_operator_review, passed_diagnostics.\n\nPredicate glossary:\n- has_field_note: has a field note\n- has_shift_clearance: has shift clearance\n- may_start_shift: may start its shift\n- needs_operator_review: needs operator review\n- passed_diagnostics: passed diagnostics\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Rover 1 has shift clearance.\n- Rover 1 passed diagnostics.\nNegative facts:\n- none\n\nTarget atom: has_field_note::Rover 6.\nTarget statement: Rover 6 has a field note.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "robotics_operations_0175_open_missing_with_closed_distractor__cwa__multi_agent", "source_id": "robotics_operations_0175_open_missing_with_closed_distractor__cwa", "base_id": "robotics_operations_0175_open_missing_with_closed_distractor", "split": "train", "domain": "robotics_operations", "family": "open_missing_with_closed_distractor", "subset": "core_contrastive", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["has_field_note", "has_shift_clearance", "may_start_shift", "needs_operator_review", "passed_diagnostics"], "agent_a_complete_predicates": ["has_field_note", "has_shift_clearance", "may_start_shift", "needs_operator_review", "passed_diagnostics"], "agent_b_open_predicates": [], "facts_positive": ["Rover 3 has shift clearance.", "Rover 3 passed diagnostics."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_diagnostics": "passed diagnostics", "has_shift_clearance": "has shift clearance", "has_field_note": "has a field note", "may_start_shift": "may start its shift", "needs_operator_review": "needs operator review"}, "vocabulary_predicates": ["has_field_note", "has_shift_clearance", "may_start_shift", "needs_operator_review", "passed_diagnostics"], "query_atom": "has_field_note::Rover 2", "target_statement": "Rover 2 has a field note.", "gold_reason_type": "global_closed_world_underivable", "symbolic": {"positive_atoms": ["has_shift_clearance::Rover 3", "passed_diagnostics::Rover 3"], "negative_atoms": [], "rules": [], "query_atom": "has_field_note::Rover 2", "closure_atoms": ["has_shift_clearance::Rover 3", "passed_diagnostics::Rover 3"]}, "source_reports": {"agent_a": {"positive": ["Rover 3 has shift clearance.", "Rover 3 passed diagnostics."], "negative": []}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "closed_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_field_note, has_shift_clearance, may_start_shift, needs_operator_review, passed_diagnostics.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- has_field_note: has a field note\n- has_shift_clearance: has shift clearance\n- may_start_shift: may start its shift\n- needs_operator_review: needs operator review\n- passed_diagnostics: passed diagnostics\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- Rover 3 has shift clearance.\n- Rover 3 passed diagnostics.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: has_field_note::Rover 2.\nTarget statement: Rover 2 has a field note.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "robotics_operations_0175_open_missing_with_closed_distractor__lcwa__multi_agent", "source_id": "robotics_operations_0175_open_missing_with_closed_distractor__lcwa", "base_id": "robotics_operations_0175_open_missing_with_closed_distractor", "split": "train", "domain": "robotics_operations", "family": "open_missing_with_closed_distractor", "subset": "core_contrastive", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_shift_clearance", "may_start_shift"], "agent_a_complete_predicates": ["has_shift_clearance", "may_start_shift"], "agent_b_open_predicates": ["has_field_note", "needs_operator_review", "passed_diagnostics"], "facts_positive": ["Rover 3 has shift clearance.", "Rover 3 passed diagnostics."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_diagnostics": "passed diagnostics", "has_shift_clearance": "has shift clearance", "has_field_note": "has a field note", "may_start_shift": "may start its shift", "needs_operator_review": "needs operator review"}, "vocabulary_predicates": ["has_field_note", "has_shift_clearance", "may_start_shift", "needs_operator_review", "passed_diagnostics"], "query_atom": "has_field_note::Rover 2", "target_statement": "Rover 2 has a field note.", "gold_reason_type": "local_open_world_underivable", "symbolic": {"positive_atoms": ["has_shift_clearance::Rover 3", "passed_diagnostics::Rover 3"], "negative_atoms": [], "rules": [], "query_atom": "has_field_note::Rover 2", "closure_atoms": ["has_shift_clearance::Rover 3", "passed_diagnostics::Rover 3"]}, "source_reports": {"agent_a": {"positive": ["Rover 3 has shift clearance."], "negative": []}, "agent_b": {"positive": ["Rover 3 passed diagnostics."], "negative": []}}, "gold_truth_value": "unknown", "gold_source_used": "agent_b", "gold_closure_handling": "open_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_shift_clearance, may_start_shift.\nAgent B open/incomplete predicates: has_field_note, needs_operator_review, passed_diagnostics.\n\nPredicate glossary:\n- has_field_note: has a field note\n- has_shift_clearance: has shift clearance\n- may_start_shift: may start its shift\n- needs_operator_review: needs operator review\n- passed_diagnostics: passed diagnostics\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- Rover 3 has shift clearance.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Rover 3 passed diagnostics.\nNegative facts:\n- none\n\nTarget atom: has_field_note::Rover 2.\nTarget statement: Rover 2 has a field note.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "robotics_operations_0175_open_missing_with_closed_distractor__owa__multi_agent", "source_id": "robotics_operations_0175_open_missing_with_closed_distractor__owa", "base_id": "robotics_operations_0175_open_missing_with_closed_distractor", "split": "train", "domain": "robotics_operations", "family": "open_missing_with_closed_distractor", "subset": "core_contrastive", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["has_field_note", "has_shift_clearance", "may_start_shift", "needs_operator_review", "passed_diagnostics"], "facts_positive": ["Rover 3 has shift clearance.", "Rover 3 passed diagnostics."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_diagnostics": "passed diagnostics", "has_shift_clearance": "has shift clearance", "has_field_note": "has a field note", "may_start_shift": "may start its shift", "needs_operator_review": "needs operator review"}, "vocabulary_predicates": ["has_field_note", "has_shift_clearance", "may_start_shift", "needs_operator_review", "passed_diagnostics"], "query_atom": "has_field_note::Rover 2", "target_statement": "Rover 2 has a field note.", "gold_reason_type": "open_world_underivable", "symbolic": {"positive_atoms": ["has_shift_clearance::Rover 3", "passed_diagnostics::Rover 3"], "negative_atoms": [], "rules": [], "query_atom": "has_field_note::Rover 2", "closure_atoms": ["has_shift_clearance::Rover 3", "passed_diagnostics::Rover 3"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Rover 3 has shift clearance.", "Rover 3 passed diagnostics."], "negative": []}}, "gold_truth_value": "unknown", "gold_source_used": "agent_b", "gold_closure_handling": "open_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: has_field_note, has_shift_clearance, may_start_shift, needs_operator_review, passed_diagnostics.\n\nPredicate glossary:\n- has_field_note: has a field note\n- has_shift_clearance: has shift clearance\n- may_start_shift: may start its shift\n- needs_operator_review: needs operator review\n- passed_diagnostics: passed diagnostics\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Rover 3 has shift clearance.\n- Rover 3 passed diagnostics.\nNegative facts:\n- none\n\nTarget atom: has_field_note::Rover 2.\nTarget statement: Rover 2 has a field note.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "robotics_operations_0179_closed_derived_missing_antecedent__cwa__multi_agent", "source_id": "robotics_operations_0179_closed_derived_missing_antecedent__cwa", "base_id": "robotics_operations_0179_closed_derived_missing_antecedent", "split": "train", "domain": "robotics_operations", "family": "closed_derived_missing_antecedent", "subset": "core_contrastive", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["has_field_note", "has_shift_clearance", "may_start_shift", "needs_operator_review", "passed_diagnostics"], "agent_a_complete_predicates": ["has_field_note", "has_shift_clearance", "may_start_shift", "needs_operator_review", "passed_diagnostics"], "agent_b_open_predicates": [], "facts_positive": ["Rover 6 passed diagnostics."], "facts_negative": [], "rules_natural": ["If a robot passed diagnostics and has shift clearance, then that robot may start its shift."], "predicate_glossary": {"passed_diagnostics": "passed diagnostics", "has_shift_clearance": "has shift clearance", "has_field_note": "has a field note", "may_start_shift": "may start its shift", "needs_operator_review": "needs operator review"}, "vocabulary_predicates": ["has_field_note", "has_shift_clearance", "may_start_shift", "needs_operator_review", "passed_diagnostics"], "query_atom": "may_start_shift::Rover 6", "target_statement": "Rover 6 may start its shift.", "gold_reason_type": "global_closed_world_underivable", "symbolic": {"positive_atoms": ["passed_diagnostics::Rover 6"], "negative_atoms": [], "rules": [{"antecedents": ["passed_diagnostics::Rover 6", "has_shift_clearance::Rover 6"], "conclusion": "may_start_shift::Rover 6", "text": "If a robot passed diagnostics and has shift clearance, then that robot may start its shift."}], "query_atom": "may_start_shift::Rover 6", "closure_atoms": ["passed_diagnostics::Rover 6"]}, "source_reports": {"agent_a": {"positive": ["Rover 6 passed diagnostics."], "negative": []}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "closed_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_field_note, has_shift_clearance, may_start_shift, needs_operator_review, passed_diagnostics.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- has_field_note: has a field note\n- has_shift_clearance: has shift clearance\n- may_start_shift: may start its shift\n- needs_operator_review: needs operator review\n- passed_diagnostics: passed diagnostics\n\nCoordinator policy rules:\n- If a robot passed diagnostics and has shift clearance, then that robot may start its shift.\n\nAgent A report:\nPositive facts:\n- Rover 6 passed diagnostics.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: may_start_shift::Rover 6.\nTarget statement: Rover 6 may start its shift.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "robotics_operations_0179_closed_derived_missing_antecedent__lcwa__multi_agent", "source_id": "robotics_operations_0179_closed_derived_missing_antecedent__lcwa", "base_id": "robotics_operations_0179_closed_derived_missing_antecedent", "split": "train", "domain": "robotics_operations", "family": "closed_derived_missing_antecedent", "subset": "core_contrastive", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_shift_clearance", "may_start_shift"], "agent_a_complete_predicates": ["has_shift_clearance", "may_start_shift"], "agent_b_open_predicates": ["has_field_note", "needs_operator_review", "passed_diagnostics"], "facts_positive": ["Rover 6 passed diagnostics."], "facts_negative": [], "rules_natural": ["If a robot passed diagnostics and has shift clearance, then that robot may start its shift."], "predicate_glossary": {"passed_diagnostics": "passed diagnostics", "has_shift_clearance": "has shift clearance", "has_field_note": "has a field note", "may_start_shift": "may start its shift", "needs_operator_review": "needs operator review"}, "vocabulary_predicates": ["has_field_note", "has_shift_clearance", "may_start_shift", "needs_operator_review", "passed_diagnostics"], "query_atom": "may_start_shift::Rover 6", "target_statement": "Rover 6 may start its shift.", "gold_reason_type": "local_closed_world_underivable", "symbolic": {"positive_atoms": ["passed_diagnostics::Rover 6"], "negative_atoms": [], "rules": [{"antecedents": ["passed_diagnostics::Rover 6", "has_shift_clearance::Rover 6"], "conclusion": "may_start_shift::Rover 6", "text": "If a robot passed diagnostics and has shift clearance, then that robot may start its shift."}], "query_atom": "may_start_shift::Rover 6", "closure_atoms": ["passed_diagnostics::Rover 6"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Rover 6 passed diagnostics."], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "closed_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_shift_clearance, may_start_shift.\nAgent B open/incomplete predicates: has_field_note, needs_operator_review, passed_diagnostics.\n\nPredicate glossary:\n- has_field_note: has a field note\n- has_shift_clearance: has shift clearance\n- may_start_shift: may start its shift\n- needs_operator_review: needs operator review\n- passed_diagnostics: passed diagnostics\n\nCoordinator policy rules:\n- If a robot passed diagnostics and has shift clearance, then that robot may start its shift.\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Rover 6 passed diagnostics.\nNegative facts:\n- none\n\nTarget atom: may_start_shift::Rover 6.\nTarget statement: Rover 6 may start its shift.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "robotics_operations_0179_closed_derived_missing_antecedent__owa__multi_agent", "source_id": "robotics_operations_0179_closed_derived_missing_antecedent__owa", "base_id": "robotics_operations_0179_closed_derived_missing_antecedent", "split": "train", "domain": "robotics_operations", "family": "closed_derived_missing_antecedent", "subset": "core_contrastive", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["has_field_note", "has_shift_clearance", "may_start_shift", "needs_operator_review", "passed_diagnostics"], "facts_positive": ["Rover 6 passed diagnostics."], "facts_negative": [], "rules_natural": ["If a robot passed diagnostics and has shift clearance, then that robot may start its shift."], "predicate_glossary": {"passed_diagnostics": "passed diagnostics", "has_shift_clearance": "has shift clearance", "has_field_note": "has a field note", "may_start_shift": "may start its shift", "needs_operator_review": "needs operator review"}, "vocabulary_predicates": ["has_field_note", "has_shift_clearance", "may_start_shift", "needs_operator_review", "passed_diagnostics"], "query_atom": "may_start_shift::Rover 6", "target_statement": "Rover 6 may start its shift.", "gold_reason_type": "open_world_underivable", "symbolic": {"positive_atoms": ["passed_diagnostics::Rover 6"], "negative_atoms": [], "rules": [{"antecedents": ["passed_diagnostics::Rover 6", "has_shift_clearance::Rover 6"], "conclusion": "may_start_shift::Rover 6", "text": "If a robot passed diagnostics and has shift clearance, then that robot may start its shift."}], "query_atom": "may_start_shift::Rover 6", "closure_atoms": ["passed_diagnostics::Rover 6"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Rover 6 passed diagnostics."], "negative": []}}, "gold_truth_value": "unknown", "gold_source_used": "agent_b", "gold_closure_handling": "open_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: has_field_note, has_shift_clearance, may_start_shift, needs_operator_review, passed_diagnostics.\n\nPredicate glossary:\n- has_field_note: has a field note\n- has_shift_clearance: has shift clearance\n- may_start_shift: may start its shift\n- needs_operator_review: needs operator review\n- passed_diagnostics: passed diagnostics\n\nCoordinator policy rules:\n- If a robot passed diagnostics and has shift clearance, then that robot may start its shift.\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Rover 6 passed diagnostics.\nNegative facts:\n- none\n\nTarget atom: may_start_shift::Rover 6.\nTarget statement: Rover 6 may start its shift.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "robotics_operations_0184_entailed_closed_conclusion__cwa__multi_agent", "source_id": "robotics_operations_0184_entailed_closed_conclusion__cwa", "base_id": "robotics_operations_0184_entailed_closed_conclusion", "split": "train", "domain": "robotics_operations", "family": "entailed_closed_conclusion", "subset": "control_entailed", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["has_field_note", "has_shift_clearance", "may_start_shift", "needs_operator_review", "passed_diagnostics"], "agent_a_complete_predicates": ["has_field_note", "has_shift_clearance", "may_start_shift", "needs_operator_review", "passed_diagnostics"], "agent_b_open_predicates": [], "facts_positive": ["Rover 5 has shift clearance.", "Rover 5 passed diagnostics."], "facts_negative": [], "rules_natural": ["If a robot passed diagnostics and has shift clearance, then that robot may start its shift."], "predicate_glossary": {"passed_diagnostics": "passed diagnostics", "has_shift_clearance": "has shift clearance", "has_field_note": "has a field note", "may_start_shift": "may start its shift", "needs_operator_review": "needs operator review"}, "vocabulary_predicates": ["has_field_note", "has_shift_clearance", "may_start_shift", "needs_operator_review", "passed_diagnostics"], "query_atom": "may_start_shift::Rover 5", "target_statement": "Rover 5 may start its shift.", "gold_reason_type": "entailed_by_fact_or_rule", "symbolic": {"positive_atoms": ["has_shift_clearance::Rover 5", "passed_diagnostics::Rover 5"], "negative_atoms": [], "rules": [{"antecedents": ["passed_diagnostics::Rover 5", "has_shift_clearance::Rover 5"], "conclusion": "may_start_shift::Rover 5", "text": "If a robot passed diagnostics and has shift clearance, then that robot may start its shift."}], "query_atom": "may_start_shift::Rover 5", "closure_atoms": ["has_shift_clearance::Rover 5", "may_start_shift::Rover 5", "passed_diagnostics::Rover 5"]}, "source_reports": {"agent_a": {"positive": ["Rover 5 has shift clearance.", "Rover 5 passed diagnostics."], "negative": []}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "true", "gold_source_used": "agent_a", "gold_closure_handling": "entailed_or_explicit", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_field_note, has_shift_clearance, may_start_shift, needs_operator_review, passed_diagnostics.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- has_field_note: has a field note\n- has_shift_clearance: has shift clearance\n- may_start_shift: may start its shift\n- needs_operator_review: needs operator review\n- passed_diagnostics: passed diagnostics\n\nCoordinator policy rules:\n- If a robot passed diagnostics and has shift clearance, then that robot may start its shift.\n\nAgent A report:\nPositive facts:\n- Rover 5 has shift clearance.\n- Rover 5 passed diagnostics.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: may_start_shift::Rover 5.\nTarget statement: Rover 5 may start its shift.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "robotics_operations_0184_entailed_closed_conclusion__lcwa__multi_agent", "source_id": "robotics_operations_0184_entailed_closed_conclusion__lcwa", "base_id": "robotics_operations_0184_entailed_closed_conclusion", "split": "train", "domain": "robotics_operations", "family": "entailed_closed_conclusion", "subset": "control_entailed", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_shift_clearance", "may_start_shift"], "agent_a_complete_predicates": ["has_shift_clearance", "may_start_shift"], "agent_b_open_predicates": ["has_field_note", "needs_operator_review", "passed_diagnostics"], "facts_positive": ["Rover 5 has shift clearance.", "Rover 5 passed diagnostics."], "facts_negative": [], "rules_natural": ["If a robot passed diagnostics and has shift clearance, then that robot may start its shift."], "predicate_glossary": {"passed_diagnostics": "passed diagnostics", "has_shift_clearance": "has shift clearance", "has_field_note": "has a field note", "may_start_shift": "may start its shift", "needs_operator_review": "needs operator review"}, "vocabulary_predicates": ["has_field_note", "has_shift_clearance", "may_start_shift", "needs_operator_review", "passed_diagnostics"], "query_atom": "may_start_shift::Rover 5", "target_statement": "Rover 5 may start its shift.", "gold_reason_type": "entailed_by_fact_or_rule", "symbolic": {"positive_atoms": ["has_shift_clearance::Rover 5", "passed_diagnostics::Rover 5"], "negative_atoms": [], "rules": [{"antecedents": ["passed_diagnostics::Rover 5", "has_shift_clearance::Rover 5"], "conclusion": "may_start_shift::Rover 5", "text": "If a robot passed diagnostics and has shift clearance, then that robot may start its shift."}], "query_atom": "may_start_shift::Rover 5", "closure_atoms": ["has_shift_clearance::Rover 5", "may_start_shift::Rover 5", "passed_diagnostics::Rover 5"]}, "source_reports": {"agent_a": {"positive": ["Rover 5 has shift clearance."], "negative": []}, "agent_b": {"positive": ["Rover 5 passed diagnostics."], "negative": []}}, "gold_truth_value": "true", "gold_source_used": "both", "gold_closure_handling": "entailed_or_explicit", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_shift_clearance, may_start_shift.\nAgent B open/incomplete predicates: has_field_note, needs_operator_review, passed_diagnostics.\n\nPredicate glossary:\n- has_field_note: has a field note\n- has_shift_clearance: has shift clearance\n- may_start_shift: may start its shift\n- needs_operator_review: needs operator review\n- passed_diagnostics: passed diagnostics\n\nCoordinator policy rules:\n- If a robot passed diagnostics and has shift clearance, then that robot may start its shift.\n\nAgent A report:\nPositive facts:\n- Rover 5 has shift clearance.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Rover 5 passed diagnostics.\nNegative facts:\n- none\n\nTarget atom: may_start_shift::Rover 5.\nTarget statement: Rover 5 may start its shift.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "robotics_operations_0184_entailed_closed_conclusion__owa__multi_agent", "source_id": "robotics_operations_0184_entailed_closed_conclusion__owa", "base_id": "robotics_operations_0184_entailed_closed_conclusion", "split": "train", "domain": "robotics_operations", "family": "entailed_closed_conclusion", "subset": "control_entailed", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["has_field_note", "has_shift_clearance", "may_start_shift", "needs_operator_review", "passed_diagnostics"], "facts_positive": ["Rover 5 has shift clearance.", "Rover 5 passed diagnostics."], "facts_negative": [], "rules_natural": ["If a robot passed diagnostics and has shift clearance, then that robot may start its shift."], "predicate_glossary": {"passed_diagnostics": "passed diagnostics", "has_shift_clearance": "has shift clearance", "has_field_note": "has a field note", "may_start_shift": "may start its shift", "needs_operator_review": "needs operator review"}, "vocabulary_predicates": ["has_field_note", "has_shift_clearance", "may_start_shift", "needs_operator_review", "passed_diagnostics"], "query_atom": "may_start_shift::Rover 5", "target_statement": "Rover 5 may start its shift.", "gold_reason_type": "entailed_by_fact_or_rule", "symbolic": {"positive_atoms": ["has_shift_clearance::Rover 5", "passed_diagnostics::Rover 5"], "negative_atoms": [], "rules": [{"antecedents": ["passed_diagnostics::Rover 5", "has_shift_clearance::Rover 5"], "conclusion": "may_start_shift::Rover 5", "text": "If a robot passed diagnostics and has shift clearance, then that robot may start its shift."}], "query_atom": "may_start_shift::Rover 5", "closure_atoms": ["has_shift_clearance::Rover 5", "may_start_shift::Rover 5", "passed_diagnostics::Rover 5"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Rover 5 has shift clearance.", "Rover 5 passed diagnostics."], "negative": []}}, "gold_truth_value": "true", "gold_source_used": "agent_b", "gold_closure_handling": "entailed_or_explicit", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: has_field_note, has_shift_clearance, may_start_shift, needs_operator_review, passed_diagnostics.\n\nPredicate glossary:\n- has_field_note: has a field note\n- has_shift_clearance: has shift clearance\n- may_start_shift: may start its shift\n- needs_operator_review: needs operator review\n- passed_diagnostics: passed diagnostics\n\nCoordinator policy rules:\n- If a robot passed diagnostics and has shift clearance, then that robot may start its shift.\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Rover 5 has shift clearance.\n- Rover 5 passed diagnostics.\nNegative facts:\n- none\n\nTarget atom: may_start_shift::Rover 5.\nTarget statement: Rover 5 may start its shift.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "robotics_operations_0185_entailed_closed_conclusion__cwa__multi_agent", "source_id": "robotics_operations_0185_entailed_closed_conclusion__cwa", "base_id": "robotics_operations_0185_entailed_closed_conclusion", "split": "train", "domain": "robotics_operations", "family": "entailed_closed_conclusion", "subset": "control_entailed", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["has_field_note", "has_shift_clearance", "may_start_shift", "needs_operator_review", "passed_diagnostics"], "agent_a_complete_predicates": ["has_field_note", "has_shift_clearance", "may_start_shift", "needs_operator_review", "passed_diagnostics"], "agent_b_open_predicates": [], "facts_positive": ["Rover 6 has shift clearance.", "Rover 6 passed diagnostics."], "facts_negative": [], "rules_natural": ["If a robot passed diagnostics and has shift clearance, then that robot may start its shift."], "predicate_glossary": {"passed_diagnostics": "passed diagnostics", "has_shift_clearance": "has shift clearance", "has_field_note": "has a field note", "may_start_shift": "may start its shift", "needs_operator_review": "needs operator review"}, "vocabulary_predicates": ["has_field_note", "has_shift_clearance", "may_start_shift", "needs_operator_review", "passed_diagnostics"], "query_atom": "may_start_shift::Rover 6", "target_statement": "Rover 6 may start its shift.", "gold_reason_type": "entailed_by_fact_or_rule", "symbolic": {"positive_atoms": ["has_shift_clearance::Rover 6", "passed_diagnostics::Rover 6"], "negative_atoms": [], "rules": [{"antecedents": ["passed_diagnostics::Rover 6", "has_shift_clearance::Rover 6"], "conclusion": "may_start_shift::Rover 6", "text": "If a robot passed diagnostics and has shift clearance, then that robot may start its shift."}], "query_atom": "may_start_shift::Rover 6", "closure_atoms": ["has_shift_clearance::Rover 6", "may_start_shift::Rover 6", "passed_diagnostics::Rover 6"]}, "source_reports": {"agent_a": {"positive": ["Rover 6 has shift clearance.", "Rover 6 passed diagnostics."], "negative": []}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "true", "gold_source_used": "agent_a", "gold_closure_handling": "entailed_or_explicit", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_field_note, has_shift_clearance, may_start_shift, needs_operator_review, passed_diagnostics.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- has_field_note: has a field note\n- has_shift_clearance: has shift clearance\n- may_start_shift: may start its shift\n- needs_operator_review: needs operator review\n- passed_diagnostics: passed diagnostics\n\nCoordinator policy rules:\n- If a robot passed diagnostics and has shift clearance, then that robot may start its shift.\n\nAgent A report:\nPositive facts:\n- Rover 6 has shift clearance.\n- Rover 6 passed diagnostics.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: may_start_shift::Rover 6.\nTarget statement: Rover 6 may start its shift.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "robotics_operations_0185_entailed_closed_conclusion__lcwa__multi_agent", "source_id": "robotics_operations_0185_entailed_closed_conclusion__lcwa", "base_id": "robotics_operations_0185_entailed_closed_conclusion", "split": "train", "domain": "robotics_operations", "family": "entailed_closed_conclusion", "subset": "control_entailed", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_shift_clearance", "may_start_shift"], "agent_a_complete_predicates": ["has_shift_clearance", "may_start_shift"], "agent_b_open_predicates": ["has_field_note", "needs_operator_review", "passed_diagnostics"], "facts_positive": ["Rover 6 has shift clearance.", "Rover 6 passed diagnostics."], "facts_negative": [], "rules_natural": ["If a robot passed diagnostics and has shift clearance, then that robot may start its shift."], "predicate_glossary": {"passed_diagnostics": "passed diagnostics", "has_shift_clearance": "has shift clearance", "has_field_note": "has a field note", "may_start_shift": "may start its shift", "needs_operator_review": "needs operator review"}, "vocabulary_predicates": ["has_field_note", "has_shift_clearance", "may_start_shift", "needs_operator_review", "passed_diagnostics"], "query_atom": "may_start_shift::Rover 6", "target_statement": "Rover 6 may start its shift.", "gold_reason_type": "entailed_by_fact_or_rule", "symbolic": {"positive_atoms": ["has_shift_clearance::Rover 6", "passed_diagnostics::Rover 6"], "negative_atoms": [], "rules": [{"antecedents": ["passed_diagnostics::Rover 6", "has_shift_clearance::Rover 6"], "conclusion": "may_start_shift::Rover 6", "text": "If a robot passed diagnostics and has shift clearance, then that robot may start its shift."}], "query_atom": "may_start_shift::Rover 6", "closure_atoms": ["has_shift_clearance::Rover 6", "may_start_shift::Rover 6", "passed_diagnostics::Rover 6"]}, "source_reports": {"agent_a": {"positive": ["Rover 6 has shift clearance."], "negative": []}, "agent_b": {"positive": ["Rover 6 passed diagnostics."], "negative": []}}, "gold_truth_value": "true", "gold_source_used": "both", "gold_closure_handling": "entailed_or_explicit", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_shift_clearance, may_start_shift.\nAgent B open/incomplete predicates: has_field_note, needs_operator_review, passed_diagnostics.\n\nPredicate glossary:\n- has_field_note: has a field note\n- has_shift_clearance: has shift clearance\n- may_start_shift: may start its shift\n- needs_operator_review: needs operator review\n- passed_diagnostics: passed diagnostics\n\nCoordinator policy rules:\n- If a robot passed diagnostics and has shift clearance, then that robot may start its shift.\n\nAgent A report:\nPositive facts:\n- Rover 6 has shift clearance.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Rover 6 passed diagnostics.\nNegative facts:\n- none\n\nTarget atom: may_start_shift::Rover 6.\nTarget statement: Rover 6 may start its shift.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "robotics_operations_0185_entailed_closed_conclusion__owa__multi_agent", "source_id": "robotics_operations_0185_entailed_closed_conclusion__owa", "base_id": "robotics_operations_0185_entailed_closed_conclusion", "split": "train", "domain": "robotics_operations", "family": "entailed_closed_conclusion", "subset": "control_entailed", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["has_field_note", "has_shift_clearance", "may_start_shift", "needs_operator_review", "passed_diagnostics"], "facts_positive": ["Rover 6 has shift clearance.", "Rover 6 passed diagnostics."], "facts_negative": [], "rules_natural": ["If a robot passed diagnostics and has shift clearance, then that robot may start its shift."], "predicate_glossary": {"passed_diagnostics": "passed diagnostics", "has_shift_clearance": "has shift clearance", "has_field_note": "has a field note", "may_start_shift": "may start its shift", "needs_operator_review": "needs operator review"}, "vocabulary_predicates": ["has_field_note", "has_shift_clearance", "may_start_shift", "needs_operator_review", "passed_diagnostics"], "query_atom": "may_start_shift::Rover 6", "target_statement": "Rover 6 may start its shift.", "gold_reason_type": "entailed_by_fact_or_rule", "symbolic": {"positive_atoms": ["has_shift_clearance::Rover 6", "passed_diagnostics::Rover 6"], "negative_atoms": [], "rules": [{"antecedents": ["passed_diagnostics::Rover 6", "has_shift_clearance::Rover 6"], "conclusion": "may_start_shift::Rover 6", "text": "If a robot passed diagnostics and has shift clearance, then that robot may start its shift."}], "query_atom": "may_start_shift::Rover 6", "closure_atoms": ["has_shift_clearance::Rover 6", "may_start_shift::Rover 6", "passed_diagnostics::Rover 6"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Rover 6 has shift clearance.", "Rover 6 passed diagnostics."], "negative": []}}, "gold_truth_value": "true", "gold_source_used": "agent_b", "gold_closure_handling": "entailed_or_explicit", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: has_field_note, has_shift_clearance, may_start_shift, needs_operator_review, passed_diagnostics.\n\nPredicate glossary:\n- has_field_note: has a field note\n- has_shift_clearance: has shift clearance\n- may_start_shift: may start its shift\n- needs_operator_review: needs operator review\n- passed_diagnostics: passed diagnostics\n\nCoordinator policy rules:\n- If a robot passed diagnostics and has shift clearance, then that robot may start its shift.\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Rover 6 has shift clearance.\n- Rover 6 passed diagnostics.\nNegative facts:\n- none\n\nTarget atom: may_start_shift::Rover 6.\nTarget statement: Rover 6 may start its shift.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "robotics_operations_0188_entailed_open_conclusion__cwa__multi_agent", "source_id": "robotics_operations_0188_entailed_open_conclusion__cwa", "base_id": "robotics_operations_0188_entailed_open_conclusion", "split": "test", "domain": "robotics_operations", "family": "entailed_open_conclusion", "subset": "control_entailed", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["has_field_note", "has_shift_clearance", "may_start_shift", "needs_operator_review", "passed_diagnostics"], "agent_a_complete_predicates": ["has_field_note", "has_shift_clearance", "may_start_shift", "needs_operator_review", "passed_diagnostics"], "agent_b_open_predicates": [], "facts_positive": ["Rover 3 has a field note."], "facts_negative": [], "rules_natural": ["If a robot has a field note, then that robot needs operator review."], "predicate_glossary": {"passed_diagnostics": "passed diagnostics", "has_shift_clearance": "has shift clearance", "has_field_note": "has a field note", "may_start_shift": "may start its shift", "needs_operator_review": "needs operator review"}, "vocabulary_predicates": ["has_field_note", "has_shift_clearance", "may_start_shift", "needs_operator_review", "passed_diagnostics"], "query_atom": "needs_operator_review::Rover 3", "target_statement": "Rover 3 needs operator review.", "gold_reason_type": "entailed_by_fact_or_rule", "symbolic": {"positive_atoms": ["has_field_note::Rover 3"], "negative_atoms": [], "rules": [{"antecedents": ["has_field_note::Rover 3"], "conclusion": "needs_operator_review::Rover 3", "text": "If a robot has a field note, then that robot needs operator review."}], "query_atom": "needs_operator_review::Rover 3", "closure_atoms": ["has_field_note::Rover 3", "needs_operator_review::Rover 3"]}, "source_reports": {"agent_a": {"positive": ["Rover 3 has a field note."], "negative": []}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "true", "gold_source_used": "agent_a", "gold_closure_handling": "entailed_or_explicit", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_field_note, has_shift_clearance, may_start_shift, needs_operator_review, passed_diagnostics.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- has_field_note: has a field note\n- has_shift_clearance: has shift clearance\n- may_start_shift: may start its shift\n- needs_operator_review: needs operator review\n- passed_diagnostics: passed diagnostics\n\nCoordinator policy rules:\n- If a robot has a field note, then that robot needs operator review.\n\nAgent A report:\nPositive facts:\n- Rover 3 has a field note.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: needs_operator_review::Rover 3.\nTarget statement: Rover 3 needs operator review.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "robotics_operations_0188_entailed_open_conclusion__lcwa__multi_agent", "source_id": "robotics_operations_0188_entailed_open_conclusion__lcwa", "base_id": "robotics_operations_0188_entailed_open_conclusion", "split": "test", "domain": "robotics_operations", "family": "entailed_open_conclusion", "subset": "control_entailed", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_shift_clearance", "may_start_shift"], "agent_a_complete_predicates": ["has_shift_clearance", "may_start_shift"], "agent_b_open_predicates": ["has_field_note", "needs_operator_review", "passed_diagnostics"], "facts_positive": ["Rover 3 has a field note."], "facts_negative": [], "rules_natural": ["If a robot has a field note, then that robot needs operator review."], "predicate_glossary": {"passed_diagnostics": "passed diagnostics", "has_shift_clearance": "has shift clearance", "has_field_note": "has a field note", "may_start_shift": "may start its shift", "needs_operator_review": "needs operator review"}, "vocabulary_predicates": ["has_field_note", "has_shift_clearance", "may_start_shift", "needs_operator_review", "passed_diagnostics"], "query_atom": "needs_operator_review::Rover 3", "target_statement": "Rover 3 needs operator review.", "gold_reason_type": "entailed_by_fact_or_rule", "symbolic": {"positive_atoms": ["has_field_note::Rover 3"], "negative_atoms": [], "rules": [{"antecedents": ["has_field_note::Rover 3"], "conclusion": "needs_operator_review::Rover 3", "text": "If a robot has a field note, then that robot needs operator review."}], "query_atom": "needs_operator_review::Rover 3", "closure_atoms": ["has_field_note::Rover 3", "needs_operator_review::Rover 3"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Rover 3 has a field note."], "negative": []}}, "gold_truth_value": "true", "gold_source_used": "agent_b", "gold_closure_handling": "entailed_or_explicit", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_shift_clearance, may_start_shift.\nAgent B open/incomplete predicates: has_field_note, needs_operator_review, passed_diagnostics.\n\nPredicate glossary:\n- has_field_note: has a field note\n- has_shift_clearance: has shift clearance\n- may_start_shift: may start its shift\n- needs_operator_review: needs operator review\n- passed_diagnostics: passed diagnostics\n\nCoordinator policy rules:\n- If a robot has a field note, then that robot needs operator review.\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Rover 3 has a field note.\nNegative facts:\n- none\n\nTarget atom: needs_operator_review::Rover 3.\nTarget statement: Rover 3 needs operator review.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "robotics_operations_0188_entailed_open_conclusion__owa__multi_agent", "source_id": "robotics_operations_0188_entailed_open_conclusion__owa", "base_id": "robotics_operations_0188_entailed_open_conclusion", "split": "test", "domain": "robotics_operations", "family": "entailed_open_conclusion", "subset": "control_entailed", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["has_field_note", "has_shift_clearance", "may_start_shift", "needs_operator_review", "passed_diagnostics"], "facts_positive": ["Rover 3 has a field note."], "facts_negative": [], "rules_natural": ["If a robot has a field note, then that robot needs operator review."], "predicate_glossary": {"passed_diagnostics": "passed diagnostics", "has_shift_clearance": "has shift clearance", "has_field_note": "has a field note", "may_start_shift": "may start its shift", "needs_operator_review": "needs operator review"}, "vocabulary_predicates": ["has_field_note", "has_shift_clearance", "may_start_shift", "needs_operator_review", "passed_diagnostics"], "query_atom": "needs_operator_review::Rover 3", "target_statement": "Rover 3 needs operator review.", "gold_reason_type": "entailed_by_fact_or_rule", "symbolic": {"positive_atoms": ["has_field_note::Rover 3"], "negative_atoms": [], "rules": [{"antecedents": ["has_field_note::Rover 3"], "conclusion": "needs_operator_review::Rover 3", "text": "If a robot has a field note, then that robot needs operator review."}], "query_atom": "needs_operator_review::Rover 3", "closure_atoms": ["has_field_note::Rover 3", "needs_operator_review::Rover 3"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Rover 3 has a field note."], "negative": []}}, "gold_truth_value": "true", "gold_source_used": "agent_b", "gold_closure_handling": "entailed_or_explicit", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: has_field_note, has_shift_clearance, may_start_shift, needs_operator_review, passed_diagnostics.\n\nPredicate glossary:\n- has_field_note: has a field note\n- has_shift_clearance: has shift clearance\n- may_start_shift: may start its shift\n- needs_operator_review: needs operator review\n- passed_diagnostics: passed diagnostics\n\nCoordinator policy rules:\n- If a robot has a field note, then that robot needs operator review.\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Rover 3 has a field note.\nNegative facts:\n- none\n\nTarget atom: needs_operator_review::Rover 3.\nTarget statement: Rover 3 needs operator review.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "robotics_operations_0190_entailed_open_conclusion__cwa__multi_agent", "source_id": "robotics_operations_0190_entailed_open_conclusion__cwa", "base_id": "robotics_operations_0190_entailed_open_conclusion", "split": "train", "domain": "robotics_operations", "family": "entailed_open_conclusion", "subset": "control_entailed", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["has_field_note", "has_shift_clearance", "may_start_shift", "needs_operator_review", "passed_diagnostics"], "agent_a_complete_predicates": ["has_field_note", "has_shift_clearance", "may_start_shift", "needs_operator_review", "passed_diagnostics"], "agent_b_open_predicates": [], "facts_positive": ["Rover 5 has a field note."], "facts_negative": [], "rules_natural": ["If a robot has a field note, then that robot needs operator review."], "predicate_glossary": {"passed_diagnostics": "passed diagnostics", "has_shift_clearance": "has shift clearance", "has_field_note": "has a field note", "may_start_shift": "may start its shift", "needs_operator_review": "needs operator review"}, "vocabulary_predicates": ["has_field_note", "has_shift_clearance", "may_start_shift", "needs_operator_review", "passed_diagnostics"], "query_atom": "needs_operator_review::Rover 5", "target_statement": "Rover 5 needs operator review.", "gold_reason_type": "entailed_by_fact_or_rule", "symbolic": {"positive_atoms": ["has_field_note::Rover 5"], "negative_atoms": [], "rules": [{"antecedents": ["has_field_note::Rover 5"], "conclusion": "needs_operator_review::Rover 5", "text": "If a robot has a field note, then that robot needs operator review."}], "query_atom": "needs_operator_review::Rover 5", "closure_atoms": ["has_field_note::Rover 5", "needs_operator_review::Rover 5"]}, "source_reports": {"agent_a": {"positive": ["Rover 5 has a field note."], "negative": []}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "true", "gold_source_used": "agent_a", "gold_closure_handling": "entailed_or_explicit", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_field_note, has_shift_clearance, may_start_shift, needs_operator_review, passed_diagnostics.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- has_field_note: has a field note\n- has_shift_clearance: has shift clearance\n- may_start_shift: may start its shift\n- needs_operator_review: needs operator review\n- passed_diagnostics: passed diagnostics\n\nCoordinator policy rules:\n- If a robot has a field note, then that robot needs operator review.\n\nAgent A report:\nPositive facts:\n- Rover 5 has a field note.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: needs_operator_review::Rover 5.\nTarget statement: Rover 5 needs operator review.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "robotics_operations_0190_entailed_open_conclusion__lcwa__multi_agent", "source_id": "robotics_operations_0190_entailed_open_conclusion__lcwa", "base_id": "robotics_operations_0190_entailed_open_conclusion", "split": "train", "domain": "robotics_operations", "family": "entailed_open_conclusion", "subset": "control_entailed", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_shift_clearance", "may_start_shift"], "agent_a_complete_predicates": ["has_shift_clearance", "may_start_shift"], "agent_b_open_predicates": ["has_field_note", "needs_operator_review", "passed_diagnostics"], "facts_positive": ["Rover 5 has a field note."], "facts_negative": [], "rules_natural": ["If a robot has a field note, then that robot needs operator review."], "predicate_glossary": {"passed_diagnostics": "passed diagnostics", "has_shift_clearance": "has shift clearance", "has_field_note": "has a field note", "may_start_shift": "may start its shift", "needs_operator_review": "needs operator review"}, "vocabulary_predicates": ["has_field_note", "has_shift_clearance", "may_start_shift", "needs_operator_review", "passed_diagnostics"], "query_atom": "needs_operator_review::Rover 5", "target_statement": "Rover 5 needs operator review.", "gold_reason_type": "entailed_by_fact_or_rule", "symbolic": {"positive_atoms": ["has_field_note::Rover 5"], "negative_atoms": [], "rules": [{"antecedents": ["has_field_note::Rover 5"], "conclusion": "needs_operator_review::Rover 5", "text": "If a robot has a field note, then that robot needs operator review."}], "query_atom": "needs_operator_review::Rover 5", "closure_atoms": ["has_field_note::Rover 5", "needs_operator_review::Rover 5"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Rover 5 has a field note."], "negative": []}}, "gold_truth_value": "true", "gold_source_used": "agent_b", "gold_closure_handling": "entailed_or_explicit", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_shift_clearance, may_start_shift.\nAgent B open/incomplete predicates: has_field_note, needs_operator_review, passed_diagnostics.\n\nPredicate glossary:\n- has_field_note: has a field note\n- has_shift_clearance: has shift clearance\n- may_start_shift: may start its shift\n- needs_operator_review: needs operator review\n- passed_diagnostics: passed diagnostics\n\nCoordinator policy rules:\n- If a robot has a field note, then that robot needs operator review.\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Rover 5 has a field note.\nNegative facts:\n- none\n\nTarget atom: needs_operator_review::Rover 5.\nTarget statement: Rover 5 needs operator review.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "robotics_operations_0190_entailed_open_conclusion__owa__multi_agent", "source_id": "robotics_operations_0190_entailed_open_conclusion__owa", "base_id": "robotics_operations_0190_entailed_open_conclusion", "split": "train", "domain": "robotics_operations", "family": "entailed_open_conclusion", "subset": "control_entailed", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["has_field_note", "has_shift_clearance", "may_start_shift", "needs_operator_review", "passed_diagnostics"], "facts_positive": ["Rover 5 has a field note."], "facts_negative": [], "rules_natural": ["If a robot has a field note, then that robot needs operator review."], "predicate_glossary": {"passed_diagnostics": "passed diagnostics", "has_shift_clearance": "has shift clearance", "has_field_note": "has a field note", "may_start_shift": "may start its shift", "needs_operator_review": "needs operator review"}, "vocabulary_predicates": ["has_field_note", "has_shift_clearance", "may_start_shift", "needs_operator_review", "passed_diagnostics"], "query_atom": "needs_operator_review::Rover 5", "target_statement": "Rover 5 needs operator review.", "gold_reason_type": "entailed_by_fact_or_rule", "symbolic": {"positive_atoms": ["has_field_note::Rover 5"], "negative_atoms": [], "rules": [{"antecedents": ["has_field_note::Rover 5"], "conclusion": "needs_operator_review::Rover 5", "text": "If a robot has a field note, then that robot needs operator review."}], "query_atom": "needs_operator_review::Rover 5", "closure_atoms": ["has_field_note::Rover 5", "needs_operator_review::Rover 5"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Rover 5 has a field note."], "negative": []}}, "gold_truth_value": "true", "gold_source_used": "agent_b", "gold_closure_handling": "entailed_or_explicit", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: has_field_note, has_shift_clearance, may_start_shift, needs_operator_review, passed_diagnostics.\n\nPredicate glossary:\n- has_field_note: has a field note\n- has_shift_clearance: has shift clearance\n- may_start_shift: may start its shift\n- needs_operator_review: needs operator review\n- passed_diagnostics: passed diagnostics\n\nCoordinator policy rules:\n- If a robot has a field note, then that robot needs operator review.\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Rover 5 has a field note.\nNegative facts:\n- none\n\nTarget atom: needs_operator_review::Rover 5.\nTarget statement: Rover 5 needs operator review.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "robotics_operations_0194_explicit_negative_closed__cwa__multi_agent", "source_id": "robotics_operations_0194_explicit_negative_closed__cwa", "base_id": "robotics_operations_0194_explicit_negative_closed", "split": "train", "domain": "robotics_operations", "family": "explicit_negative_closed", "subset": "control_explicit_negative", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["has_field_note", "has_shift_clearance", "may_start_shift", "needs_operator_review", "passed_diagnostics"], "agent_a_complete_predicates": ["has_field_note", "has_shift_clearance", "may_start_shift", "needs_operator_review", "passed_diagnostics"], "agent_b_open_predicates": [], "facts_positive": ["Rover 5 passed diagnostics."], "facts_negative": ["Rover 3 does not has shift clearance."], "rules_natural": [], "predicate_glossary": {"passed_diagnostics": "passed diagnostics", "has_shift_clearance": "has shift clearance", "has_field_note": "has a field note", "may_start_shift": "may start its shift", "needs_operator_review": "needs operator review"}, "vocabulary_predicates": ["has_field_note", "has_shift_clearance", "may_start_shift", "needs_operator_review", "passed_diagnostics"], "query_atom": "has_shift_clearance::Rover 3", "target_statement": "Rover 3 has shift clearance.", "gold_reason_type": "explicit_negative_fact", "symbolic": {"positive_atoms": ["passed_diagnostics::Rover 5"], "negative_atoms": ["has_shift_clearance::Rover 3"], "rules": [], "query_atom": "has_shift_clearance::Rover 3", "closure_atoms": ["passed_diagnostics::Rover 5"]}, "source_reports": {"agent_a": {"positive": ["Rover 5 passed diagnostics."], "negative": ["Rover 3 does not has shift clearance."]}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "explicit_negative", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_field_note, has_shift_clearance, may_start_shift, needs_operator_review, passed_diagnostics.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- has_field_note: has a field note\n- has_shift_clearance: has shift clearance\n- may_start_shift: may start its shift\n- needs_operator_review: needs operator review\n- passed_diagnostics: passed diagnostics\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- Rover 5 passed diagnostics.\nNegative facts:\n- Rover 3 does not has shift clearance.\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: has_shift_clearance::Rover 3.\nTarget statement: Rover 3 has shift clearance.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "robotics_operations_0194_explicit_negative_closed__lcwa__multi_agent", "source_id": "robotics_operations_0194_explicit_negative_closed__lcwa", "base_id": "robotics_operations_0194_explicit_negative_closed", "split": "train", "domain": "robotics_operations", "family": "explicit_negative_closed", "subset": "control_explicit_negative", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_shift_clearance"], "agent_a_complete_predicates": ["has_shift_clearance"], "agent_b_open_predicates": ["has_field_note", "may_start_shift", "needs_operator_review", "passed_diagnostics"], "facts_positive": ["Rover 5 passed diagnostics."], "facts_negative": ["Rover 3 does not has shift clearance."], "rules_natural": [], "predicate_glossary": {"passed_diagnostics": "passed diagnostics", "has_shift_clearance": "has shift clearance", "has_field_note": "has a field note", "may_start_shift": "may start its shift", "needs_operator_review": "needs operator review"}, "vocabulary_predicates": ["has_field_note", "has_shift_clearance", "may_start_shift", "needs_operator_review", "passed_diagnostics"], "query_atom": "has_shift_clearance::Rover 3", "target_statement": "Rover 3 has shift clearance.", "gold_reason_type": "explicit_negative_fact", "symbolic": {"positive_atoms": ["passed_diagnostics::Rover 5"], "negative_atoms": ["has_shift_clearance::Rover 3"], "rules": [], "query_atom": "has_shift_clearance::Rover 3", "closure_atoms": ["passed_diagnostics::Rover 5"]}, "source_reports": {"agent_a": {"positive": [], "negative": ["Rover 3 does not has shift clearance."]}, "agent_b": {"positive": ["Rover 5 passed diagnostics."], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "explicit_negative", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_shift_clearance.\nAgent B open/incomplete predicates: has_field_note, may_start_shift, needs_operator_review, passed_diagnostics.\n\nPredicate glossary:\n- has_field_note: has a field note\n- has_shift_clearance: has shift clearance\n- may_start_shift: may start its shift\n- needs_operator_review: needs operator review\n- passed_diagnostics: passed diagnostics\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- Rover 3 does not has shift clearance.\n\nAgent B report:\nPositive facts:\n- Rover 5 passed diagnostics.\nNegative facts:\n- none\n\nTarget atom: has_shift_clearance::Rover 3.\nTarget statement: Rover 3 has shift clearance.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "robotics_operations_0194_explicit_negative_closed__owa__multi_agent", "source_id": "robotics_operations_0194_explicit_negative_closed__owa", "base_id": "robotics_operations_0194_explicit_negative_closed", "split": "train", "domain": "robotics_operations", "family": "explicit_negative_closed", "subset": "control_explicit_negative", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["has_field_note", "has_shift_clearance", "may_start_shift", "needs_operator_review", "passed_diagnostics"], "facts_positive": ["Rover 5 passed diagnostics."], "facts_negative": ["Rover 3 does not has shift clearance."], "rules_natural": [], "predicate_glossary": {"passed_diagnostics": "passed diagnostics", "has_shift_clearance": "has shift clearance", "has_field_note": "has a field note", "may_start_shift": "may start its shift", "needs_operator_review": "needs operator review"}, "vocabulary_predicates": ["has_field_note", "has_shift_clearance", "may_start_shift", "needs_operator_review", "passed_diagnostics"], "query_atom": "has_shift_clearance::Rover 3", "target_statement": "Rover 3 has shift clearance.", "gold_reason_type": "explicit_negative_fact", "symbolic": {"positive_atoms": ["passed_diagnostics::Rover 5"], "negative_atoms": ["has_shift_clearance::Rover 3"], "rules": [], "query_atom": "has_shift_clearance::Rover 3", "closure_atoms": ["passed_diagnostics::Rover 5"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Rover 5 passed diagnostics."], "negative": ["Rover 3 does not has shift clearance."]}}, "gold_truth_value": "false", "gold_source_used": "agent_b", "gold_closure_handling": "explicit_negative", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: has_field_note, has_shift_clearance, may_start_shift, needs_operator_review, passed_diagnostics.\n\nPredicate glossary:\n- has_field_note: has a field note\n- has_shift_clearance: has shift clearance\n- may_start_shift: may start its shift\n- needs_operator_review: needs operator review\n- passed_diagnostics: passed diagnostics\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Rover 5 passed diagnostics.\nNegative facts:\n- Rover 3 does not has shift clearance.\n\nTarget atom: has_shift_clearance::Rover 3.\nTarget statement: Rover 3 has shift clearance.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "robotics_operations_0196_explicit_negative_open__cwa__multi_agent", "source_id": "robotics_operations_0196_explicit_negative_open__cwa", "base_id": "robotics_operations_0196_explicit_negative_open", "split": "train", "domain": "robotics_operations", "family": "explicit_negative_open", "subset": "control_explicit_negative", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["has_field_note", "has_shift_clearance", "may_start_shift", "needs_operator_review", "passed_diagnostics"], "agent_a_complete_predicates": ["has_field_note", "has_shift_clearance", "may_start_shift", "needs_operator_review", "passed_diagnostics"], "agent_b_open_predicates": [], "facts_positive": ["Rover 1 passed diagnostics."], "facts_negative": ["Rover 5 does not has a field note."], "rules_natural": [], "predicate_glossary": {"passed_diagnostics": "passed diagnostics", "has_shift_clearance": "has shift clearance", "has_field_note": "has a field note", "may_start_shift": "may start its shift", "needs_operator_review": "needs operator review"}, "vocabulary_predicates": ["has_field_note", "has_shift_clearance", "may_start_shift", "needs_operator_review", "passed_diagnostics"], "query_atom": "has_field_note::Rover 5", "target_statement": "Rover 5 has a field note.", "gold_reason_type": "explicit_negative_fact", "symbolic": {"positive_atoms": ["passed_diagnostics::Rover 1"], "negative_atoms": ["has_field_note::Rover 5"], "rules": [], "query_atom": "has_field_note::Rover 5", "closure_atoms": ["passed_diagnostics::Rover 1"]}, "source_reports": {"agent_a": {"positive": ["Rover 1 passed diagnostics."], "negative": ["Rover 5 does not has a field note."]}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "explicit_negative", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_field_note, has_shift_clearance, may_start_shift, needs_operator_review, passed_diagnostics.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- has_field_note: has a field note\n- has_shift_clearance: has shift clearance\n- may_start_shift: may start its shift\n- needs_operator_review: needs operator review\n- passed_diagnostics: passed diagnostics\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- Rover 1 passed diagnostics.\nNegative facts:\n- Rover 5 does not has a field note.\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: has_field_note::Rover 5.\nTarget statement: Rover 5 has a field note.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "robotics_operations_0196_explicit_negative_open__lcwa__multi_agent", "source_id": "robotics_operations_0196_explicit_negative_open__lcwa", "base_id": "robotics_operations_0196_explicit_negative_open", "split": "train", "domain": "robotics_operations", "family": "explicit_negative_open", "subset": "control_explicit_negative", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_shift_clearance", "may_start_shift"], "agent_a_complete_predicates": ["has_shift_clearance", "may_start_shift"], "agent_b_open_predicates": ["has_field_note", "needs_operator_review", "passed_diagnostics"], "facts_positive": ["Rover 1 passed diagnostics."], "facts_negative": ["Rover 5 does not has a field note."], "rules_natural": [], "predicate_glossary": {"passed_diagnostics": "passed diagnostics", "has_shift_clearance": "has shift clearance", "has_field_note": "has a field note", "may_start_shift": "may start its shift", "needs_operator_review": "needs operator review"}, "vocabulary_predicates": ["has_field_note", "has_shift_clearance", "may_start_shift", "needs_operator_review", "passed_diagnostics"], "query_atom": "has_field_note::Rover 5", "target_statement": "Rover 5 has a field note.", "gold_reason_type": "explicit_negative_fact", "symbolic": {"positive_atoms": ["passed_diagnostics::Rover 1"], "negative_atoms": ["has_field_note::Rover 5"], "rules": [], "query_atom": "has_field_note::Rover 5", "closure_atoms": ["passed_diagnostics::Rover 1"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Rover 1 passed diagnostics."], "negative": ["Rover 5 does not has a field note."]}}, "gold_truth_value": "false", "gold_source_used": "agent_b", "gold_closure_handling": "explicit_negative", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_shift_clearance, may_start_shift.\nAgent B open/incomplete predicates: has_field_note, needs_operator_review, passed_diagnostics.\n\nPredicate glossary:\n- has_field_note: has a field note\n- has_shift_clearance: has shift clearance\n- may_start_shift: may start its shift\n- needs_operator_review: needs operator review\n- passed_diagnostics: passed diagnostics\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Rover 1 passed diagnostics.\nNegative facts:\n- Rover 5 does not has a field note.\n\nTarget atom: has_field_note::Rover 5.\nTarget statement: Rover 5 has a field note.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "robotics_operations_0196_explicit_negative_open__owa__multi_agent", "source_id": "robotics_operations_0196_explicit_negative_open__owa", "base_id": "robotics_operations_0196_explicit_negative_open", "split": "train", "domain": "robotics_operations", "family": "explicit_negative_open", "subset": "control_explicit_negative", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["has_field_note", "has_shift_clearance", "may_start_shift", "needs_operator_review", "passed_diagnostics"], "facts_positive": ["Rover 1 passed diagnostics."], "facts_negative": ["Rover 5 does not has a field note."], "rules_natural": [], "predicate_glossary": {"passed_diagnostics": "passed diagnostics", "has_shift_clearance": "has shift clearance", "has_field_note": "has a field note", "may_start_shift": "may start its shift", "needs_operator_review": "needs operator review"}, "vocabulary_predicates": ["has_field_note", "has_shift_clearance", "may_start_shift", "needs_operator_review", "passed_diagnostics"], "query_atom": "has_field_note::Rover 5", "target_statement": "Rover 5 has a field note.", "gold_reason_type": "explicit_negative_fact", "symbolic": {"positive_atoms": ["passed_diagnostics::Rover 1"], "negative_atoms": ["has_field_note::Rover 5"], "rules": [], "query_atom": "has_field_note::Rover 5", "closure_atoms": ["passed_diagnostics::Rover 1"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Rover 1 passed diagnostics."], "negative": ["Rover 5 does not has a field note."]}}, "gold_truth_value": "false", "gold_source_used": "agent_b", "gold_closure_handling": "explicit_negative", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: has_field_note, has_shift_clearance, may_start_shift, needs_operator_review, passed_diagnostics.\n\nPredicate glossary:\n- has_field_note: has a field note\n- has_shift_clearance: has shift clearance\n- may_start_shift: may start its shift\n- needs_operator_review: needs operator review\n- passed_diagnostics: passed diagnostics\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Rover 1 passed diagnostics.\nNegative facts:\n- Rover 5 does not has a field note.\n\nTarget atom: has_field_note::Rover 5.\nTarget statement: Rover 5 has a field note.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "robotics_operations_0198_explicit_negative_open__cwa__multi_agent", "source_id": "robotics_operations_0198_explicit_negative_open__cwa", "base_id": "robotics_operations_0198_explicit_negative_open", "split": "train", "domain": "robotics_operations", "family": "explicit_negative_open", "subset": "control_explicit_negative", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["has_field_note", "has_shift_clearance", "may_start_shift", "needs_operator_review", "passed_diagnostics"], "agent_a_complete_predicates": ["has_field_note", "has_shift_clearance", "may_start_shift", "needs_operator_review", "passed_diagnostics"], "agent_b_open_predicates": [], "facts_positive": ["Rover 3 passed diagnostics."], "facts_negative": ["Rover 1 does not has a field note."], "rules_natural": [], "predicate_glossary": {"passed_diagnostics": "passed diagnostics", "has_shift_clearance": "has shift clearance", "has_field_note": "has a field note", "may_start_shift": "may start its shift", "needs_operator_review": "needs operator review"}, "vocabulary_predicates": ["has_field_note", "has_shift_clearance", "may_start_shift", "needs_operator_review", "passed_diagnostics"], "query_atom": "has_field_note::Rover 1", "target_statement": "Rover 1 has a field note.", "gold_reason_type": "explicit_negative_fact", "symbolic": {"positive_atoms": ["passed_diagnostics::Rover 3"], "negative_atoms": ["has_field_note::Rover 1"], "rules": [], "query_atom": "has_field_note::Rover 1", "closure_atoms": ["passed_diagnostics::Rover 3"]}, "source_reports": {"agent_a": {"positive": ["Rover 3 passed diagnostics."], "negative": ["Rover 1 does not has a field note."]}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "explicit_negative", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_field_note, has_shift_clearance, may_start_shift, needs_operator_review, passed_diagnostics.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- has_field_note: has a field note\n- has_shift_clearance: has shift clearance\n- may_start_shift: may start its shift\n- needs_operator_review: needs operator review\n- passed_diagnostics: passed diagnostics\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- Rover 3 passed diagnostics.\nNegative facts:\n- Rover 1 does not has a field note.\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: has_field_note::Rover 1.\nTarget statement: Rover 1 has a field note.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "robotics_operations_0198_explicit_negative_open__lcwa__multi_agent", "source_id": "robotics_operations_0198_explicit_negative_open__lcwa", "base_id": "robotics_operations_0198_explicit_negative_open", "split": "train", "domain": "robotics_operations", "family": "explicit_negative_open", "subset": "control_explicit_negative", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_shift_clearance", "may_start_shift"], "agent_a_complete_predicates": ["has_shift_clearance", "may_start_shift"], "agent_b_open_predicates": ["has_field_note", "needs_operator_review", "passed_diagnostics"], "facts_positive": ["Rover 3 passed diagnostics."], "facts_negative": ["Rover 1 does not has a field note."], "rules_natural": [], "predicate_glossary": {"passed_diagnostics": "passed diagnostics", "has_shift_clearance": "has shift clearance", "has_field_note": "has a field note", "may_start_shift": "may start its shift", "needs_operator_review": "needs operator review"}, "vocabulary_predicates": ["has_field_note", "has_shift_clearance", "may_start_shift", "needs_operator_review", "passed_diagnostics"], "query_atom": "has_field_note::Rover 1", "target_statement": "Rover 1 has a field note.", "gold_reason_type": "explicit_negative_fact", "symbolic": {"positive_atoms": ["passed_diagnostics::Rover 3"], "negative_atoms": ["has_field_note::Rover 1"], "rules": [], "query_atom": "has_field_note::Rover 1", "closure_atoms": ["passed_diagnostics::Rover 3"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Rover 3 passed diagnostics."], "negative": ["Rover 1 does not has a field note."]}}, "gold_truth_value": "false", "gold_source_used": "agent_b", "gold_closure_handling": "explicit_negative", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_shift_clearance, may_start_shift.\nAgent B open/incomplete predicates: has_field_note, needs_operator_review, passed_diagnostics.\n\nPredicate glossary:\n- has_field_note: has a field note\n- has_shift_clearance: has shift clearance\n- may_start_shift: may start its shift\n- needs_operator_review: needs operator review\n- passed_diagnostics: passed diagnostics\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Rover 3 passed diagnostics.\nNegative facts:\n- Rover 1 does not has a field note.\n\nTarget atom: has_field_note::Rover 1.\nTarget statement: Rover 1 has a field note.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "robotics_operations_0198_explicit_negative_open__owa__multi_agent", "source_id": "robotics_operations_0198_explicit_negative_open__owa", "base_id": "robotics_operations_0198_explicit_negative_open", "split": "train", "domain": "robotics_operations", "family": "explicit_negative_open", "subset": "control_explicit_negative", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["has_field_note", "has_shift_clearance", "may_start_shift", "needs_operator_review", "passed_diagnostics"], "facts_positive": ["Rover 3 passed diagnostics."], "facts_negative": ["Rover 1 does not has a field note."], "rules_natural": [], "predicate_glossary": {"passed_diagnostics": "passed diagnostics", "has_shift_clearance": "has shift clearance", "has_field_note": "has a field note", "may_start_shift": "may start its shift", "needs_operator_review": "needs operator review"}, "vocabulary_predicates": ["has_field_note", "has_shift_clearance", "may_start_shift", "needs_operator_review", "passed_diagnostics"], "query_atom": "has_field_note::Rover 1", "target_statement": "Rover 1 has a field note.", "gold_reason_type": "explicit_negative_fact", "symbolic": {"positive_atoms": ["passed_diagnostics::Rover 3"], "negative_atoms": ["has_field_note::Rover 1"], "rules": [], "query_atom": "has_field_note::Rover 1", "closure_atoms": ["passed_diagnostics::Rover 3"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Rover 3 passed diagnostics."], "negative": ["Rover 1 does not has a field note."]}}, "gold_truth_value": "false", "gold_source_used": "agent_b", "gold_closure_handling": "explicit_negative", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: has_field_note, has_shift_clearance, may_start_shift, needs_operator_review, passed_diagnostics.\n\nPredicate glossary:\n- has_field_note: has a field note\n- has_shift_clearance: has shift clearance\n- may_start_shift: may start its shift\n- needs_operator_review: needs operator review\n- passed_diagnostics: passed diagnostics\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Rover 3 passed diagnostics.\nNegative facts:\n- Rover 1 does not has a field note.\n\nTarget atom: has_field_note::Rover 1.\nTarget statement: Rover 1 has a field note.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "robotics_operations_0199_explicit_negative_open__cwa__multi_agent", "source_id": "robotics_operations_0199_explicit_negative_open__cwa", "base_id": "robotics_operations_0199_explicit_negative_open", "split": "train", "domain": "robotics_operations", "family": "explicit_negative_open", "subset": "control_explicit_negative", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["has_field_note", "has_shift_clearance", "may_start_shift", "needs_operator_review", "passed_diagnostics"], "agent_a_complete_predicates": ["has_field_note", "has_shift_clearance", "may_start_shift", "needs_operator_review", "passed_diagnostics"], "agent_b_open_predicates": [], "facts_positive": ["Rover 4 passed diagnostics."], "facts_negative": ["Rover 2 does not has a field note."], "rules_natural": [], "predicate_glossary": {"passed_diagnostics": "passed diagnostics", "has_shift_clearance": "has shift clearance", "has_field_note": "has a field note", "may_start_shift": "may start its shift", "needs_operator_review": "needs operator review"}, "vocabulary_predicates": ["has_field_note", "has_shift_clearance", "may_start_shift", "needs_operator_review", "passed_diagnostics"], "query_atom": "has_field_note::Rover 2", "target_statement": "Rover 2 has a field note.", "gold_reason_type": "explicit_negative_fact", "symbolic": {"positive_atoms": ["passed_diagnostics::Rover 4"], "negative_atoms": ["has_field_note::Rover 2"], "rules": [], "query_atom": "has_field_note::Rover 2", "closure_atoms": ["passed_diagnostics::Rover 4"]}, "source_reports": {"agent_a": {"positive": ["Rover 4 passed diagnostics."], "negative": ["Rover 2 does not has a field note."]}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "explicit_negative", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_field_note, has_shift_clearance, may_start_shift, needs_operator_review, passed_diagnostics.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- has_field_note: has a field note\n- has_shift_clearance: has shift clearance\n- may_start_shift: may start its shift\n- needs_operator_review: needs operator review\n- passed_diagnostics: passed diagnostics\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- Rover 4 passed diagnostics.\nNegative facts:\n- Rover 2 does not has a field note.\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: has_field_note::Rover 2.\nTarget statement: Rover 2 has a field note.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "robotics_operations_0199_explicit_negative_open__lcwa__multi_agent", "source_id": "robotics_operations_0199_explicit_negative_open__lcwa", "base_id": "robotics_operations_0199_explicit_negative_open", "split": "train", "domain": "robotics_operations", "family": "explicit_negative_open", "subset": "control_explicit_negative", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_shift_clearance", "may_start_shift"], "agent_a_complete_predicates": ["has_shift_clearance", "may_start_shift"], "agent_b_open_predicates": ["has_field_note", "needs_operator_review", "passed_diagnostics"], "facts_positive": ["Rover 4 passed diagnostics."], "facts_negative": ["Rover 2 does not has a field note."], "rules_natural": [], "predicate_glossary": {"passed_diagnostics": "passed diagnostics", "has_shift_clearance": "has shift clearance", "has_field_note": "has a field note", "may_start_shift": "may start its shift", "needs_operator_review": "needs operator review"}, "vocabulary_predicates": ["has_field_note", "has_shift_clearance", "may_start_shift", "needs_operator_review", "passed_diagnostics"], "query_atom": "has_field_note::Rover 2", "target_statement": "Rover 2 has a field note.", "gold_reason_type": "explicit_negative_fact", "symbolic": {"positive_atoms": ["passed_diagnostics::Rover 4"], "negative_atoms": ["has_field_note::Rover 2"], "rules": [], "query_atom": "has_field_note::Rover 2", "closure_atoms": ["passed_diagnostics::Rover 4"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Rover 4 passed diagnostics."], "negative": ["Rover 2 does not has a field note."]}}, "gold_truth_value": "false", "gold_source_used": "agent_b", "gold_closure_handling": "explicit_negative", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_shift_clearance, may_start_shift.\nAgent B open/incomplete predicates: has_field_note, needs_operator_review, passed_diagnostics.\n\nPredicate glossary:\n- has_field_note: has a field note\n- has_shift_clearance: has shift clearance\n- may_start_shift: may start its shift\n- needs_operator_review: needs operator review\n- passed_diagnostics: passed diagnostics\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Rover 4 passed diagnostics.\nNegative facts:\n- Rover 2 does not has a field note.\n\nTarget atom: has_field_note::Rover 2.\nTarget statement: Rover 2 has a field note.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "robotics_operations_0199_explicit_negative_open__owa__multi_agent", "source_id": "robotics_operations_0199_explicit_negative_open__owa", "base_id": "robotics_operations_0199_explicit_negative_open", "split": "train", "domain": "robotics_operations", "family": "explicit_negative_open", "subset": "control_explicit_negative", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["has_field_note", "has_shift_clearance", "may_start_shift", "needs_operator_review", "passed_diagnostics"], "facts_positive": ["Rover 4 passed diagnostics."], "facts_negative": ["Rover 2 does not has a field note."], "rules_natural": [], "predicate_glossary": {"passed_diagnostics": "passed diagnostics", "has_shift_clearance": "has shift clearance", "has_field_note": "has a field note", "may_start_shift": "may start its shift", "needs_operator_review": "needs operator review"}, "vocabulary_predicates": ["has_field_note", "has_shift_clearance", "may_start_shift", "needs_operator_review", "passed_diagnostics"], "query_atom": "has_field_note::Rover 2", "target_statement": "Rover 2 has a field note.", "gold_reason_type": "explicit_negative_fact", "symbolic": {"positive_atoms": ["passed_diagnostics::Rover 4"], "negative_atoms": ["has_field_note::Rover 2"], "rules": [], "query_atom": "has_field_note::Rover 2", "closure_atoms": ["passed_diagnostics::Rover 4"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Rover 4 passed diagnostics."], "negative": ["Rover 2 does not has a field note."]}}, "gold_truth_value": "false", "gold_source_used": "agent_b", "gold_closure_handling": "explicit_negative", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: has_field_note, has_shift_clearance, may_start_shift, needs_operator_review, passed_diagnostics.\n\nPredicate glossary:\n- has_field_note: has a field note\n- has_shift_clearance: has shift clearance\n- may_start_shift: may start its shift\n- needs_operator_review: needs operator review\n- passed_diagnostics: passed diagnostics\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Rover 4 passed diagnostics.\nNegative facts:\n- Rover 2 does not has a field note.\n\nTarget atom: has_field_note::Rover 2.\nTarget statement: Rover 2 has a field note.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "university_registration_0041_closed_missing_direct__cwa__multi_agent", "source_id": "university_registration_0041_closed_missing_direct__cwa", "base_id": "university_registration_0041_closed_missing_direct", "split": "train", "domain": "university_registration", "family": "closed_missing_direct", "subset": "core_contrastive", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["has_advising_clearance", "has_external_scholarship", "may_register", "needs_manual_audit", "passed_exam"], "agent_a_complete_predicates": ["has_advising_clearance", "has_external_scholarship", "may_register", "needs_manual_audit", "passed_exam"], "agent_b_open_predicates": [], "facts_positive": ["Theo passed the qualifying exam."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_exam": "passed the qualifying exam", "has_advising_clearance": "has advising clearance", "has_external_scholarship": "has an external scholarship", "may_register": "may register for the seminar", "needs_manual_audit": "needs a manual audit"}, "vocabulary_predicates": ["has_advising_clearance", "has_external_scholarship", "may_register", "needs_manual_audit", "passed_exam"], "query_atom": "has_advising_clearance::Theo", "target_statement": "Theo has advising clearance.", "gold_reason_type": "global_closed_world_underivable", "symbolic": {"positive_atoms": ["passed_exam::Theo"], "negative_atoms": [], "rules": [], "query_atom": "has_advising_clearance::Theo", "closure_atoms": ["passed_exam::Theo"]}, "source_reports": {"agent_a": {"positive": ["Theo passed the qualifying exam."], "negative": []}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "closed_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_advising_clearance, has_external_scholarship, may_register, needs_manual_audit, passed_exam.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- has_advising_clearance: has advising clearance\n- has_external_scholarship: has an external scholarship\n- may_register: may register for the seminar\n- needs_manual_audit: needs a manual audit\n- passed_exam: passed the qualifying exam\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- Theo passed the qualifying exam.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: has_advising_clearance::Theo.\nTarget statement: Theo has advising clearance.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "university_registration_0041_closed_missing_direct__lcwa__multi_agent", "source_id": "university_registration_0041_closed_missing_direct__lcwa", "base_id": "university_registration_0041_closed_missing_direct", "split": "train", "domain": "university_registration", "family": "closed_missing_direct", "subset": "core_contrastive", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_advising_clearance"], "agent_a_complete_predicates": ["has_advising_clearance"], "agent_b_open_predicates": ["has_external_scholarship", "may_register", "needs_manual_audit", "passed_exam"], "facts_positive": ["Theo passed the qualifying exam."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_exam": "passed the qualifying exam", "has_advising_clearance": "has advising clearance", "has_external_scholarship": "has an external scholarship", "may_register": "may register for the seminar", "needs_manual_audit": "needs a manual audit"}, "vocabulary_predicates": ["has_advising_clearance", "has_external_scholarship", "may_register", "needs_manual_audit", "passed_exam"], "query_atom": "has_advising_clearance::Theo", "target_statement": "Theo has advising clearance.", "gold_reason_type": "local_closed_world_underivable", "symbolic": {"positive_atoms": ["passed_exam::Theo"], "negative_atoms": [], "rules": [], "query_atom": "has_advising_clearance::Theo", "closure_atoms": ["passed_exam::Theo"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Theo passed the qualifying exam."], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "closed_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_advising_clearance.\nAgent B open/incomplete predicates: has_external_scholarship, may_register, needs_manual_audit, passed_exam.\n\nPredicate glossary:\n- has_advising_clearance: has advising clearance\n- has_external_scholarship: has an external scholarship\n- may_register: may register for the seminar\n- needs_manual_audit: needs a manual audit\n- passed_exam: passed the qualifying exam\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Theo passed the qualifying exam.\nNegative facts:\n- none\n\nTarget atom: has_advising_clearance::Theo.\nTarget statement: Theo has advising clearance.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "university_registration_0041_closed_missing_direct__owa__multi_agent", "source_id": "university_registration_0041_closed_missing_direct__owa", "base_id": "university_registration_0041_closed_missing_direct", "split": "train", "domain": "university_registration", "family": "closed_missing_direct", "subset": "core_contrastive", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["has_advising_clearance", "has_external_scholarship", "may_register", "needs_manual_audit", "passed_exam"], "facts_positive": ["Theo passed the qualifying exam."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_exam": "passed the qualifying exam", "has_advising_clearance": "has advising clearance", "has_external_scholarship": "has an external scholarship", "may_register": "may register for the seminar", "needs_manual_audit": "needs a manual audit"}, "vocabulary_predicates": ["has_advising_clearance", "has_external_scholarship", "may_register", "needs_manual_audit", "passed_exam"], "query_atom": "has_advising_clearance::Theo", "target_statement": "Theo has advising clearance.", "gold_reason_type": "open_world_underivable", "symbolic": {"positive_atoms": ["passed_exam::Theo"], "negative_atoms": [], "rules": [], "query_atom": "has_advising_clearance::Theo", "closure_atoms": ["passed_exam::Theo"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Theo passed the qualifying exam."], "negative": []}}, "gold_truth_value": "unknown", "gold_source_used": "agent_b", "gold_closure_handling": "open_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: has_advising_clearance, has_external_scholarship, may_register, needs_manual_audit, passed_exam.\n\nPredicate glossary:\n- has_advising_clearance: has advising clearance\n- has_external_scholarship: has an external scholarship\n- may_register: may register for the seminar\n- needs_manual_audit: needs a manual audit\n- passed_exam: passed the qualifying exam\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Theo passed the qualifying exam.\nNegative facts:\n- none\n\nTarget atom: has_advising_clearance::Theo.\nTarget statement: Theo has advising clearance.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "university_registration_0044_open_missing_direct__cwa__multi_agent", "source_id": "university_registration_0044_open_missing_direct__cwa", "base_id": "university_registration_0044_open_missing_direct", "split": "train", "domain": "university_registration", "family": "open_missing_direct", "subset": "core_contrastive", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["has_advising_clearance", "has_external_scholarship", "may_register", "needs_manual_audit", "passed_exam"], "agent_a_complete_predicates": ["has_advising_clearance", "has_external_scholarship", "may_register", "needs_manual_audit", "passed_exam"], "agent_b_open_predicates": [], "facts_positive": ["Lena passed the qualifying exam."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_exam": "passed the qualifying exam", "has_advising_clearance": "has advising clearance", "has_external_scholarship": "has an external scholarship", "may_register": "may register for the seminar", "needs_manual_audit": "needs a manual audit"}, "vocabulary_predicates": ["has_advising_clearance", "has_external_scholarship", "may_register", "needs_manual_audit", "passed_exam"], "query_atom": "has_external_scholarship::Lena", "target_statement": "Lena has an external scholarship.", "gold_reason_type": "global_closed_world_underivable", "symbolic": {"positive_atoms": ["passed_exam::Lena"], "negative_atoms": [], "rules": [], "query_atom": "has_external_scholarship::Lena", "closure_atoms": ["passed_exam::Lena"]}, "source_reports": {"agent_a": {"positive": ["Lena passed the qualifying exam."], "negative": []}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "closed_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_advising_clearance, has_external_scholarship, may_register, needs_manual_audit, passed_exam.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- has_advising_clearance: has advising clearance\n- has_external_scholarship: has an external scholarship\n- may_register: may register for the seminar\n- needs_manual_audit: needs a manual audit\n- passed_exam: passed the qualifying exam\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- Lena passed the qualifying exam.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: has_external_scholarship::Lena.\nTarget statement: Lena has an external scholarship.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "university_registration_0044_open_missing_direct__lcwa__multi_agent", "source_id": "university_registration_0044_open_missing_direct__lcwa", "base_id": "university_registration_0044_open_missing_direct", "split": "train", "domain": "university_registration", "family": "open_missing_direct", "subset": "core_contrastive", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_advising_clearance", "may_register"], "agent_a_complete_predicates": ["has_advising_clearance", "may_register"], "agent_b_open_predicates": ["has_external_scholarship", "needs_manual_audit", "passed_exam"], "facts_positive": ["Lena passed the qualifying exam."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_exam": "passed the qualifying exam", "has_advising_clearance": "has advising clearance", "has_external_scholarship": "has an external scholarship", "may_register": "may register for the seminar", "needs_manual_audit": "needs a manual audit"}, "vocabulary_predicates": ["has_advising_clearance", "has_external_scholarship", "may_register", "needs_manual_audit", "passed_exam"], "query_atom": "has_external_scholarship::Lena", "target_statement": "Lena has an external scholarship.", "gold_reason_type": "local_open_world_underivable", "symbolic": {"positive_atoms": ["passed_exam::Lena"], "negative_atoms": [], "rules": [], "query_atom": "has_external_scholarship::Lena", "closure_atoms": ["passed_exam::Lena"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Lena passed the qualifying exam."], "negative": []}}, "gold_truth_value": "unknown", "gold_source_used": "agent_b", "gold_closure_handling": "open_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_advising_clearance, may_register.\nAgent B open/incomplete predicates: has_external_scholarship, needs_manual_audit, passed_exam.\n\nPredicate glossary:\n- has_advising_clearance: has advising clearance\n- has_external_scholarship: has an external scholarship\n- may_register: may register for the seminar\n- needs_manual_audit: needs a manual audit\n- passed_exam: passed the qualifying exam\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Lena passed the qualifying exam.\nNegative facts:\n- none\n\nTarget atom: has_external_scholarship::Lena.\nTarget statement: Lena has an external scholarship.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "university_registration_0044_open_missing_direct__owa__multi_agent", "source_id": "university_registration_0044_open_missing_direct__owa", "base_id": "university_registration_0044_open_missing_direct", "split": "train", "domain": "university_registration", "family": "open_missing_direct", "subset": "core_contrastive", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["has_advising_clearance", "has_external_scholarship", "may_register", "needs_manual_audit", "passed_exam"], "facts_positive": ["Lena passed the qualifying exam."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_exam": "passed the qualifying exam", "has_advising_clearance": "has advising clearance", "has_external_scholarship": "has an external scholarship", "may_register": "may register for the seminar", "needs_manual_audit": "needs a manual audit"}, "vocabulary_predicates": ["has_advising_clearance", "has_external_scholarship", "may_register", "needs_manual_audit", "passed_exam"], "query_atom": "has_external_scholarship::Lena", "target_statement": "Lena has an external scholarship.", "gold_reason_type": "open_world_underivable", "symbolic": {"positive_atoms": ["passed_exam::Lena"], "negative_atoms": [], "rules": [], "query_atom": "has_external_scholarship::Lena", "closure_atoms": ["passed_exam::Lena"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Lena passed the qualifying exam."], "negative": []}}, "gold_truth_value": "unknown", "gold_source_used": "agent_b", "gold_closure_handling": "open_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: has_advising_clearance, has_external_scholarship, may_register, needs_manual_audit, passed_exam.\n\nPredicate glossary:\n- has_advising_clearance: has advising clearance\n- has_external_scholarship: has an external scholarship\n- may_register: may register for the seminar\n- needs_manual_audit: needs a manual audit\n- passed_exam: passed the qualifying exam\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Lena passed the qualifying exam.\nNegative facts:\n- none\n\nTarget atom: has_external_scholarship::Lena.\nTarget statement: Lena has an external scholarship.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "university_registration_0045_open_missing_direct__cwa__multi_agent", "source_id": "university_registration_0045_open_missing_direct__cwa", "base_id": "university_registration_0045_open_missing_direct", "split": "test", "domain": "university_registration", "family": "open_missing_direct", "subset": "core_contrastive", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["has_advising_clearance", "has_external_scholarship", "may_register", "needs_manual_audit", "passed_exam"], "agent_a_complete_predicates": ["has_advising_clearance", "has_external_scholarship", "may_register", "needs_manual_audit", "passed_exam"], "agent_b_open_predicates": [], "facts_positive": ["Omar passed the qualifying exam."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_exam": "passed the qualifying exam", "has_advising_clearance": "has advising clearance", "has_external_scholarship": "has an external scholarship", "may_register": "may register for the seminar", "needs_manual_audit": "needs a manual audit"}, "vocabulary_predicates": ["has_advising_clearance", "has_external_scholarship", "may_register", "needs_manual_audit", "passed_exam"], "query_atom": "has_external_scholarship::Omar", "target_statement": "Omar has an external scholarship.", "gold_reason_type": "global_closed_world_underivable", "symbolic": {"positive_atoms": ["passed_exam::Omar"], "negative_atoms": [], "rules": [], "query_atom": "has_external_scholarship::Omar", "closure_atoms": ["passed_exam::Omar"]}, "source_reports": {"agent_a": {"positive": ["Omar passed the qualifying exam."], "negative": []}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "closed_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_advising_clearance, has_external_scholarship, may_register, needs_manual_audit, passed_exam.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- has_advising_clearance: has advising clearance\n- has_external_scholarship: has an external scholarship\n- may_register: may register for the seminar\n- needs_manual_audit: needs a manual audit\n- passed_exam: passed the qualifying exam\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- Omar passed the qualifying exam.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: has_external_scholarship::Omar.\nTarget statement: Omar has an external scholarship.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "university_registration_0045_open_missing_direct__lcwa__multi_agent", "source_id": "university_registration_0045_open_missing_direct__lcwa", "base_id": "university_registration_0045_open_missing_direct", "split": "test", "domain": "university_registration", "family": "open_missing_direct", "subset": "core_contrastive", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_advising_clearance", "may_register"], "agent_a_complete_predicates": ["has_advising_clearance", "may_register"], "agent_b_open_predicates": ["has_external_scholarship", "needs_manual_audit", "passed_exam"], "facts_positive": ["Omar passed the qualifying exam."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_exam": "passed the qualifying exam", "has_advising_clearance": "has advising clearance", "has_external_scholarship": "has an external scholarship", "may_register": "may register for the seminar", "needs_manual_audit": "needs a manual audit"}, "vocabulary_predicates": ["has_advising_clearance", "has_external_scholarship", "may_register", "needs_manual_audit", "passed_exam"], "query_atom": "has_external_scholarship::Omar", "target_statement": "Omar has an external scholarship.", "gold_reason_type": "local_open_world_underivable", "symbolic": {"positive_atoms": ["passed_exam::Omar"], "negative_atoms": [], "rules": [], "query_atom": "has_external_scholarship::Omar", "closure_atoms": ["passed_exam::Omar"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Omar passed the qualifying exam."], "negative": []}}, "gold_truth_value": "unknown", "gold_source_used": "agent_b", "gold_closure_handling": "open_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_advising_clearance, may_register.\nAgent B open/incomplete predicates: has_external_scholarship, needs_manual_audit, passed_exam.\n\nPredicate glossary:\n- has_advising_clearance: has advising clearance\n- has_external_scholarship: has an external scholarship\n- may_register: may register for the seminar\n- needs_manual_audit: needs a manual audit\n- passed_exam: passed the qualifying exam\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Omar passed the qualifying exam.\nNegative facts:\n- none\n\nTarget atom: has_external_scholarship::Omar.\nTarget statement: Omar has an external scholarship.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "university_registration_0045_open_missing_direct__owa__multi_agent", "source_id": "university_registration_0045_open_missing_direct__owa", "base_id": "university_registration_0045_open_missing_direct", "split": "test", "domain": "university_registration", "family": "open_missing_direct", "subset": "core_contrastive", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["has_advising_clearance", "has_external_scholarship", "may_register", "needs_manual_audit", "passed_exam"], "facts_positive": ["Omar passed the qualifying exam."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_exam": "passed the qualifying exam", "has_advising_clearance": "has advising clearance", "has_external_scholarship": "has an external scholarship", "may_register": "may register for the seminar", "needs_manual_audit": "needs a manual audit"}, "vocabulary_predicates": ["has_advising_clearance", "has_external_scholarship", "may_register", "needs_manual_audit", "passed_exam"], "query_atom": "has_external_scholarship::Omar", "target_statement": "Omar has an external scholarship.", "gold_reason_type": "open_world_underivable", "symbolic": {"positive_atoms": ["passed_exam::Omar"], "negative_atoms": [], "rules": [], "query_atom": "has_external_scholarship::Omar", "closure_atoms": ["passed_exam::Omar"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Omar passed the qualifying exam."], "negative": []}}, "gold_truth_value": "unknown", "gold_source_used": "agent_b", "gold_closure_handling": "open_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: has_advising_clearance, has_external_scholarship, may_register, needs_manual_audit, passed_exam.\n\nPredicate glossary:\n- has_advising_clearance: has advising clearance\n- has_external_scholarship: has an external scholarship\n- may_register: may register for the seminar\n- needs_manual_audit: needs a manual audit\n- passed_exam: passed the qualifying exam\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Omar passed the qualifying exam.\nNegative facts:\n- none\n\nTarget atom: has_external_scholarship::Omar.\nTarget statement: Omar has an external scholarship.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "university_registration_0049_closed_missing_with_open_distractor__cwa__multi_agent", "source_id": "university_registration_0049_closed_missing_with_open_distractor__cwa", "base_id": "university_registration_0049_closed_missing_with_open_distractor", "split": "train", "domain": "university_registration", "family": "closed_missing_with_open_distractor", "subset": "core_contrastive", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["has_advising_clearance", "has_external_scholarship", "may_register", "needs_manual_audit", "passed_exam"], "agent_a_complete_predicates": ["has_advising_clearance", "has_external_scholarship", "may_register", "needs_manual_audit", "passed_exam"], "agent_b_open_predicates": [], "facts_positive": ["Lena has an external scholarship.", "Noah passed the qualifying exam."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_exam": "passed the qualifying exam", "has_advising_clearance": "has advising clearance", "has_external_scholarship": "has an external scholarship", "may_register": "may register for the seminar", "needs_manual_audit": "needs a manual audit"}, "vocabulary_predicates": ["has_advising_clearance", "has_external_scholarship", "may_register", "needs_manual_audit", "passed_exam"], "query_atom": "has_advising_clearance::Noah", "target_statement": "Noah has advising clearance.", "gold_reason_type": "global_closed_world_underivable", "symbolic": {"positive_atoms": ["has_external_scholarship::Lena", "passed_exam::Noah"], "negative_atoms": [], "rules": [], "query_atom": "has_advising_clearance::Noah", "closure_atoms": ["has_external_scholarship::Lena", "passed_exam::Noah"]}, "source_reports": {"agent_a": {"positive": ["Lena has an external scholarship.", "Noah passed the qualifying exam."], "negative": []}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "closed_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_advising_clearance, has_external_scholarship, may_register, needs_manual_audit, passed_exam.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- has_advising_clearance: has advising clearance\n- has_external_scholarship: has an external scholarship\n- may_register: may register for the seminar\n- needs_manual_audit: needs a manual audit\n- passed_exam: passed the qualifying exam\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- Lena has an external scholarship.\n- Noah passed the qualifying exam.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: has_advising_clearance::Noah.\nTarget statement: Noah has advising clearance.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "university_registration_0049_closed_missing_with_open_distractor__lcwa__multi_agent", "source_id": "university_registration_0049_closed_missing_with_open_distractor__lcwa", "base_id": "university_registration_0049_closed_missing_with_open_distractor", "split": "train", "domain": "university_registration", "family": "closed_missing_with_open_distractor", "subset": "core_contrastive", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_advising_clearance"], "agent_a_complete_predicates": ["has_advising_clearance"], "agent_b_open_predicates": ["has_external_scholarship", "may_register", "needs_manual_audit", "passed_exam"], "facts_positive": ["Lena has an external scholarship.", "Noah passed the qualifying exam."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_exam": "passed the qualifying exam", "has_advising_clearance": "has advising clearance", "has_external_scholarship": "has an external scholarship", "may_register": "may register for the seminar", "needs_manual_audit": "needs a manual audit"}, "vocabulary_predicates": ["has_advising_clearance", "has_external_scholarship", "may_register", "needs_manual_audit", "passed_exam"], "query_atom": "has_advising_clearance::Noah", "target_statement": "Noah has advising clearance.", "gold_reason_type": "local_closed_world_underivable", "symbolic": {"positive_atoms": ["has_external_scholarship::Lena", "passed_exam::Noah"], "negative_atoms": [], "rules": [], "query_atom": "has_advising_clearance::Noah", "closure_atoms": ["has_external_scholarship::Lena", "passed_exam::Noah"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Lena has an external scholarship.", "Noah passed the qualifying exam."], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "closed_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_advising_clearance.\nAgent B open/incomplete predicates: has_external_scholarship, may_register, needs_manual_audit, passed_exam.\n\nPredicate glossary:\n- has_advising_clearance: has advising clearance\n- has_external_scholarship: has an external scholarship\n- may_register: may register for the seminar\n- needs_manual_audit: needs a manual audit\n- passed_exam: passed the qualifying exam\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Lena has an external scholarship.\n- Noah passed the qualifying exam.\nNegative facts:\n- none\n\nTarget atom: has_advising_clearance::Noah.\nTarget statement: Noah has advising clearance.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "university_registration_0049_closed_missing_with_open_distractor__owa__multi_agent", "source_id": "university_registration_0049_closed_missing_with_open_distractor__owa", "base_id": "university_registration_0049_closed_missing_with_open_distractor", "split": "train", "domain": "university_registration", "family": "closed_missing_with_open_distractor", "subset": "core_contrastive", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["has_advising_clearance", "has_external_scholarship", "may_register", "needs_manual_audit", "passed_exam"], "facts_positive": ["Lena has an external scholarship.", "Noah passed the qualifying exam."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_exam": "passed the qualifying exam", "has_advising_clearance": "has advising clearance", "has_external_scholarship": "has an external scholarship", "may_register": "may register for the seminar", "needs_manual_audit": "needs a manual audit"}, "vocabulary_predicates": ["has_advising_clearance", "has_external_scholarship", "may_register", "needs_manual_audit", "passed_exam"], "query_atom": "has_advising_clearance::Noah", "target_statement": "Noah has advising clearance.", "gold_reason_type": "open_world_underivable", "symbolic": {"positive_atoms": ["has_external_scholarship::Lena", "passed_exam::Noah"], "negative_atoms": [], "rules": [], "query_atom": "has_advising_clearance::Noah", "closure_atoms": ["has_external_scholarship::Lena", "passed_exam::Noah"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Lena has an external scholarship.", "Noah passed the qualifying exam."], "negative": []}}, "gold_truth_value": "unknown", "gold_source_used": "agent_b", "gold_closure_handling": "open_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: has_advising_clearance, has_external_scholarship, may_register, needs_manual_audit, passed_exam.\n\nPredicate glossary:\n- has_advising_clearance: has advising clearance\n- has_external_scholarship: has an external scholarship\n- may_register: may register for the seminar\n- needs_manual_audit: needs a manual audit\n- passed_exam: passed the qualifying exam\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Lena has an external scholarship.\n- Noah passed the qualifying exam.\nNegative facts:\n- none\n\nTarget atom: has_advising_clearance::Noah.\nTarget statement: Noah has advising clearance.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "university_registration_0050_closed_missing_with_open_distractor__cwa__multi_agent", "source_id": "university_registration_0050_closed_missing_with_open_distractor__cwa", "base_id": "university_registration_0050_closed_missing_with_open_distractor", "split": "test", "domain": "university_registration", "family": "closed_missing_with_open_distractor", "subset": "core_contrastive", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["has_advising_clearance", "has_external_scholarship", "may_register", "needs_manual_audit", "passed_exam"], "agent_a_complete_predicates": ["has_advising_clearance", "has_external_scholarship", "may_register", "needs_manual_audit", "passed_exam"], "agent_b_open_predicates": [], "facts_positive": ["Omar has an external scholarship.", "Lena passed the qualifying exam."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_exam": "passed the qualifying exam", "has_advising_clearance": "has advising clearance", "has_external_scholarship": "has an external scholarship", "may_register": "may register for the seminar", "needs_manual_audit": "needs a manual audit"}, "vocabulary_predicates": ["has_advising_clearance", "has_external_scholarship", "may_register", "needs_manual_audit", "passed_exam"], "query_atom": "has_advising_clearance::Lena", "target_statement": "Lena has advising clearance.", "gold_reason_type": "global_closed_world_underivable", "symbolic": {"positive_atoms": ["has_external_scholarship::Omar", "passed_exam::Lena"], "negative_atoms": [], "rules": [], "query_atom": "has_advising_clearance::Lena", "closure_atoms": ["has_external_scholarship::Omar", "passed_exam::Lena"]}, "source_reports": {"agent_a": {"positive": ["Omar has an external scholarship.", "Lena passed the qualifying exam."], "negative": []}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "closed_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_advising_clearance, has_external_scholarship, may_register, needs_manual_audit, passed_exam.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- has_advising_clearance: has advising clearance\n- has_external_scholarship: has an external scholarship\n- may_register: may register for the seminar\n- needs_manual_audit: needs a manual audit\n- passed_exam: passed the qualifying exam\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- Omar has an external scholarship.\n- Lena passed the qualifying exam.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: has_advising_clearance::Lena.\nTarget statement: Lena has advising clearance.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "university_registration_0050_closed_missing_with_open_distractor__lcwa__multi_agent", "source_id": "university_registration_0050_closed_missing_with_open_distractor__lcwa", "base_id": "university_registration_0050_closed_missing_with_open_distractor", "split": "test", "domain": "university_registration", "family": "closed_missing_with_open_distractor", "subset": "core_contrastive", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_advising_clearance"], "agent_a_complete_predicates": ["has_advising_clearance"], "agent_b_open_predicates": ["has_external_scholarship", "may_register", "needs_manual_audit", "passed_exam"], "facts_positive": ["Omar has an external scholarship.", "Lena passed the qualifying exam."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_exam": "passed the qualifying exam", "has_advising_clearance": "has advising clearance", "has_external_scholarship": "has an external scholarship", "may_register": "may register for the seminar", "needs_manual_audit": "needs a manual audit"}, "vocabulary_predicates": ["has_advising_clearance", "has_external_scholarship", "may_register", "needs_manual_audit", "passed_exam"], "query_atom": "has_advising_clearance::Lena", "target_statement": "Lena has advising clearance.", "gold_reason_type": "local_closed_world_underivable", "symbolic": {"positive_atoms": ["has_external_scholarship::Omar", "passed_exam::Lena"], "negative_atoms": [], "rules": [], "query_atom": "has_advising_clearance::Lena", "closure_atoms": ["has_external_scholarship::Omar", "passed_exam::Lena"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Omar has an external scholarship.", "Lena passed the qualifying exam."], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "closed_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_advising_clearance.\nAgent B open/incomplete predicates: has_external_scholarship, may_register, needs_manual_audit, passed_exam.\n\nPredicate glossary:\n- has_advising_clearance: has advising clearance\n- has_external_scholarship: has an external scholarship\n- may_register: may register for the seminar\n- needs_manual_audit: needs a manual audit\n- passed_exam: passed the qualifying exam\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Omar has an external scholarship.\n- Lena passed the qualifying exam.\nNegative facts:\n- none\n\nTarget atom: has_advising_clearance::Lena.\nTarget statement: Lena has advising clearance.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "university_registration_0050_closed_missing_with_open_distractor__owa__multi_agent", "source_id": "university_registration_0050_closed_missing_with_open_distractor__owa", "base_id": "university_registration_0050_closed_missing_with_open_distractor", "split": "test", "domain": "university_registration", "family": "closed_missing_with_open_distractor", "subset": "core_contrastive", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["has_advising_clearance", "has_external_scholarship", "may_register", "needs_manual_audit", "passed_exam"], "facts_positive": ["Omar has an external scholarship.", "Lena passed the qualifying exam."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_exam": "passed the qualifying exam", "has_advising_clearance": "has advising clearance", "has_external_scholarship": "has an external scholarship", "may_register": "may register for the seminar", "needs_manual_audit": "needs a manual audit"}, "vocabulary_predicates": ["has_advising_clearance", "has_external_scholarship", "may_register", "needs_manual_audit", "passed_exam"], "query_atom": "has_advising_clearance::Lena", "target_statement": "Lena has advising clearance.", "gold_reason_type": "open_world_underivable", "symbolic": {"positive_atoms": ["has_external_scholarship::Omar", "passed_exam::Lena"], "negative_atoms": [], "rules": [], "query_atom": "has_advising_clearance::Lena", "closure_atoms": ["has_external_scholarship::Omar", "passed_exam::Lena"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Omar has an external scholarship.", "Lena passed the qualifying exam."], "negative": []}}, "gold_truth_value": "unknown", "gold_source_used": "agent_b", "gold_closure_handling": "open_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: has_advising_clearance, has_external_scholarship, may_register, needs_manual_audit, passed_exam.\n\nPredicate glossary:\n- has_advising_clearance: has advising clearance\n- has_external_scholarship: has an external scholarship\n- may_register: may register for the seminar\n- needs_manual_audit: needs a manual audit\n- passed_exam: passed the qualifying exam\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Omar has an external scholarship.\n- Lena passed the qualifying exam.\nNegative facts:\n- none\n\nTarget atom: has_advising_clearance::Lena.\nTarget statement: Lena has advising clearance.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "university_registration_0052_open_missing_with_closed_distractor__cwa__multi_agent", "source_id": "university_registration_0052_open_missing_with_closed_distractor__cwa", "base_id": "university_registration_0052_open_missing_with_closed_distractor", "split": "test", "domain": "university_registration", "family": "open_missing_with_closed_distractor", "subset": "core_contrastive", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["has_advising_clearance", "has_external_scholarship", "may_register", "needs_manual_audit", "passed_exam"], "agent_a_complete_predicates": ["has_advising_clearance", "has_external_scholarship", "may_register", "needs_manual_audit", "passed_exam"], "agent_b_open_predicates": [], "facts_positive": ["Theo has advising clearance.", "Theo passed the qualifying exam."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_exam": "passed the qualifying exam", "has_advising_clearance": "has advising clearance", "has_external_scholarship": "has an external scholarship", "may_register": "may register for the seminar", "needs_manual_audit": "needs a manual audit"}, "vocabulary_predicates": ["has_advising_clearance", "has_external_scholarship", "may_register", "needs_manual_audit", "passed_exam"], "query_atom": "has_external_scholarship::Iris", "target_statement": "Iris has an external scholarship.", "gold_reason_type": "global_closed_world_underivable", "symbolic": {"positive_atoms": ["has_advising_clearance::Theo", "passed_exam::Theo"], "negative_atoms": [], "rules": [], "query_atom": "has_external_scholarship::Iris", "closure_atoms": ["has_advising_clearance::Theo", "passed_exam::Theo"]}, "source_reports": {"agent_a": {"positive": ["Theo has advising clearance.", "Theo passed the qualifying exam."], "negative": []}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "closed_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_advising_clearance, has_external_scholarship, may_register, needs_manual_audit, passed_exam.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- has_advising_clearance: has advising clearance\n- has_external_scholarship: has an external scholarship\n- may_register: may register for the seminar\n- needs_manual_audit: needs a manual audit\n- passed_exam: passed the qualifying exam\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- Theo has advising clearance.\n- Theo passed the qualifying exam.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: has_external_scholarship::Iris.\nTarget statement: Iris has an external scholarship.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "university_registration_0052_open_missing_with_closed_distractor__lcwa__multi_agent", "source_id": "university_registration_0052_open_missing_with_closed_distractor__lcwa", "base_id": "university_registration_0052_open_missing_with_closed_distractor", "split": "test", "domain": "university_registration", "family": "open_missing_with_closed_distractor", "subset": "core_contrastive", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_advising_clearance", "may_register"], "agent_a_complete_predicates": ["has_advising_clearance", "may_register"], "agent_b_open_predicates": ["has_external_scholarship", "needs_manual_audit", "passed_exam"], "facts_positive": ["Theo has advising clearance.", "Theo passed the qualifying exam."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_exam": "passed the qualifying exam", "has_advising_clearance": "has advising clearance", "has_external_scholarship": "has an external scholarship", "may_register": "may register for the seminar", "needs_manual_audit": "needs a manual audit"}, "vocabulary_predicates": ["has_advising_clearance", "has_external_scholarship", "may_register", "needs_manual_audit", "passed_exam"], "query_atom": "has_external_scholarship::Iris", "target_statement": "Iris has an external scholarship.", "gold_reason_type": "local_open_world_underivable", "symbolic": {"positive_atoms": ["has_advising_clearance::Theo", "passed_exam::Theo"], "negative_atoms": [], "rules": [], "query_atom": "has_external_scholarship::Iris", "closure_atoms": ["has_advising_clearance::Theo", "passed_exam::Theo"]}, "source_reports": {"agent_a": {"positive": ["Theo has advising clearance."], "negative": []}, "agent_b": {"positive": ["Theo passed the qualifying exam."], "negative": []}}, "gold_truth_value": "unknown", "gold_source_used": "agent_b", "gold_closure_handling": "open_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_advising_clearance, may_register.\nAgent B open/incomplete predicates: has_external_scholarship, needs_manual_audit, passed_exam.\n\nPredicate glossary:\n- has_advising_clearance: has advising clearance\n- has_external_scholarship: has an external scholarship\n- may_register: may register for the seminar\n- needs_manual_audit: needs a manual audit\n- passed_exam: passed the qualifying exam\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- Theo has advising clearance.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Theo passed the qualifying exam.\nNegative facts:\n- none\n\nTarget atom: has_external_scholarship::Iris.\nTarget statement: Iris has an external scholarship.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "university_registration_0052_open_missing_with_closed_distractor__owa__multi_agent", "source_id": "university_registration_0052_open_missing_with_closed_distractor__owa", "base_id": "university_registration_0052_open_missing_with_closed_distractor", "split": "test", "domain": "university_registration", "family": "open_missing_with_closed_distractor", "subset": "core_contrastive", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["has_advising_clearance", "has_external_scholarship", "may_register", "needs_manual_audit", "passed_exam"], "facts_positive": ["Theo has advising clearance.", "Theo passed the qualifying exam."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_exam": "passed the qualifying exam", "has_advising_clearance": "has advising clearance", "has_external_scholarship": "has an external scholarship", "may_register": "may register for the seminar", "needs_manual_audit": "needs a manual audit"}, "vocabulary_predicates": ["has_advising_clearance", "has_external_scholarship", "may_register", "needs_manual_audit", "passed_exam"], "query_atom": "has_external_scholarship::Iris", "target_statement": "Iris has an external scholarship.", "gold_reason_type": "open_world_underivable", "symbolic": {"positive_atoms": ["has_advising_clearance::Theo", "passed_exam::Theo"], "negative_atoms": [], "rules": [], "query_atom": "has_external_scholarship::Iris", "closure_atoms": ["has_advising_clearance::Theo", "passed_exam::Theo"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Theo has advising clearance.", "Theo passed the qualifying exam."], "negative": []}}, "gold_truth_value": "unknown", "gold_source_used": "agent_b", "gold_closure_handling": "open_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: has_advising_clearance, has_external_scholarship, may_register, needs_manual_audit, passed_exam.\n\nPredicate glossary:\n- has_advising_clearance: has advising clearance\n- has_external_scholarship: has an external scholarship\n- may_register: may register for the seminar\n- needs_manual_audit: needs a manual audit\n- passed_exam: passed the qualifying exam\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Theo has advising clearance.\n- Theo passed the qualifying exam.\nNegative facts:\n- none\n\nTarget atom: has_external_scholarship::Iris.\nTarget statement: Iris has an external scholarship.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "university_registration_0055_open_missing_with_closed_distractor__cwa__multi_agent", "source_id": "university_registration_0055_open_missing_with_closed_distractor__cwa", "base_id": "university_registration_0055_open_missing_with_closed_distractor", "split": "train", "domain": "university_registration", "family": "open_missing_with_closed_distractor", "subset": "core_contrastive", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["has_advising_clearance", "has_external_scholarship", "may_register", "needs_manual_audit", "passed_exam"], "agent_a_complete_predicates": ["has_advising_clearance", "has_external_scholarship", "may_register", "needs_manual_audit", "passed_exam"], "agent_b_open_predicates": [], "facts_positive": ["Lena has advising clearance.", "Lena passed the qualifying exam."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_exam": "passed the qualifying exam", "has_advising_clearance": "has advising clearance", "has_external_scholarship": "has an external scholarship", "may_register": "may register for the seminar", "needs_manual_audit": "needs a manual audit"}, "vocabulary_predicates": ["has_advising_clearance", "has_external_scholarship", "may_register", "needs_manual_audit", "passed_exam"], "query_atom": "has_external_scholarship::Noah", "target_statement": "Noah has an external scholarship.", "gold_reason_type": "global_closed_world_underivable", "symbolic": {"positive_atoms": ["has_advising_clearance::Lena", "passed_exam::Lena"], "negative_atoms": [], "rules": [], "query_atom": "has_external_scholarship::Noah", "closure_atoms": ["has_advising_clearance::Lena", "passed_exam::Lena"]}, "source_reports": {"agent_a": {"positive": ["Lena has advising clearance.", "Lena passed the qualifying exam."], "negative": []}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "closed_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_advising_clearance, has_external_scholarship, may_register, needs_manual_audit, passed_exam.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- has_advising_clearance: has advising clearance\n- has_external_scholarship: has an external scholarship\n- may_register: may register for the seminar\n- needs_manual_audit: needs a manual audit\n- passed_exam: passed the qualifying exam\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- Lena has advising clearance.\n- Lena passed the qualifying exam.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: has_external_scholarship::Noah.\nTarget statement: Noah has an external scholarship.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "university_registration_0055_open_missing_with_closed_distractor__lcwa__multi_agent", "source_id": "university_registration_0055_open_missing_with_closed_distractor__lcwa", "base_id": "university_registration_0055_open_missing_with_closed_distractor", "split": "train", "domain": "university_registration", "family": "open_missing_with_closed_distractor", "subset": "core_contrastive", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_advising_clearance", "may_register"], "agent_a_complete_predicates": ["has_advising_clearance", "may_register"], "agent_b_open_predicates": ["has_external_scholarship", "needs_manual_audit", "passed_exam"], "facts_positive": ["Lena has advising clearance.", "Lena passed the qualifying exam."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_exam": "passed the qualifying exam", "has_advising_clearance": "has advising clearance", "has_external_scholarship": "has an external scholarship", "may_register": "may register for the seminar", "needs_manual_audit": "needs a manual audit"}, "vocabulary_predicates": ["has_advising_clearance", "has_external_scholarship", "may_register", "needs_manual_audit", "passed_exam"], "query_atom": "has_external_scholarship::Noah", "target_statement": "Noah has an external scholarship.", "gold_reason_type": "local_open_world_underivable", "symbolic": {"positive_atoms": ["has_advising_clearance::Lena", "passed_exam::Lena"], "negative_atoms": [], "rules": [], "query_atom": "has_external_scholarship::Noah", "closure_atoms": ["has_advising_clearance::Lena", "passed_exam::Lena"]}, "source_reports": {"agent_a": {"positive": ["Lena has advising clearance."], "negative": []}, "agent_b": {"positive": ["Lena passed the qualifying exam."], "negative": []}}, "gold_truth_value": "unknown", "gold_source_used": "agent_b", "gold_closure_handling": "open_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_advising_clearance, may_register.\nAgent B open/incomplete predicates: has_external_scholarship, needs_manual_audit, passed_exam.\n\nPredicate glossary:\n- has_advising_clearance: has advising clearance\n- has_external_scholarship: has an external scholarship\n- may_register: may register for the seminar\n- needs_manual_audit: needs a manual audit\n- passed_exam: passed the qualifying exam\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- Lena has advising clearance.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Lena passed the qualifying exam.\nNegative facts:\n- none\n\nTarget atom: has_external_scholarship::Noah.\nTarget statement: Noah has an external scholarship.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "university_registration_0055_open_missing_with_closed_distractor__owa__multi_agent", "source_id": "university_registration_0055_open_missing_with_closed_distractor__owa", "base_id": "university_registration_0055_open_missing_with_closed_distractor", "split": "train", "domain": "university_registration", "family": "open_missing_with_closed_distractor", "subset": "core_contrastive", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["has_advising_clearance", "has_external_scholarship", "may_register", "needs_manual_audit", "passed_exam"], "facts_positive": ["Lena has advising clearance.", "Lena passed the qualifying exam."], "facts_negative": [], "rules_natural": [], "predicate_glossary": {"passed_exam": "passed the qualifying exam", "has_advising_clearance": "has advising clearance", "has_external_scholarship": "has an external scholarship", "may_register": "may register for the seminar", "needs_manual_audit": "needs a manual audit"}, "vocabulary_predicates": ["has_advising_clearance", "has_external_scholarship", "may_register", "needs_manual_audit", "passed_exam"], "query_atom": "has_external_scholarship::Noah", "target_statement": "Noah has an external scholarship.", "gold_reason_type": "open_world_underivable", "symbolic": {"positive_atoms": ["has_advising_clearance::Lena", "passed_exam::Lena"], "negative_atoms": [], "rules": [], "query_atom": "has_external_scholarship::Noah", "closure_atoms": ["has_advising_clearance::Lena", "passed_exam::Lena"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Lena has advising clearance.", "Lena passed the qualifying exam."], "negative": []}}, "gold_truth_value": "unknown", "gold_source_used": "agent_b", "gold_closure_handling": "open_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: has_advising_clearance, has_external_scholarship, may_register, needs_manual_audit, passed_exam.\n\nPredicate glossary:\n- has_advising_clearance: has advising clearance\n- has_external_scholarship: has an external scholarship\n- may_register: may register for the seminar\n- needs_manual_audit: needs a manual audit\n- passed_exam: passed the qualifying exam\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Lena has advising clearance.\n- Lena passed the qualifying exam.\nNegative facts:\n- none\n\nTarget atom: has_external_scholarship::Noah.\nTarget statement: Noah has an external scholarship.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "university_registration_0056_closed_derived_missing_antecedent__cwa__multi_agent", "source_id": "university_registration_0056_closed_derived_missing_antecedent__cwa", "base_id": "university_registration_0056_closed_derived_missing_antecedent", "split": "dev", "domain": "university_registration", "family": "closed_derived_missing_antecedent", "subset": "core_contrastive", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["has_advising_clearance", "has_external_scholarship", "may_register", "needs_manual_audit", "passed_exam"], "agent_a_complete_predicates": ["has_advising_clearance", "has_external_scholarship", "may_register", "needs_manual_audit", "passed_exam"], "agent_b_open_predicates": [], "facts_positive": ["Lena passed the qualifying exam."], "facts_negative": [], "rules_natural": ["If a student passed the qualifying exam and has advising clearance, then that student may register for the seminar."], "predicate_glossary": {"passed_exam": "passed the qualifying exam", "has_advising_clearance": "has advising clearance", "has_external_scholarship": "has an external scholarship", "may_register": "may register for the seminar", "needs_manual_audit": "needs a manual audit"}, "vocabulary_predicates": ["has_advising_clearance", "has_external_scholarship", "may_register", "needs_manual_audit", "passed_exam"], "query_atom": "may_register::Lena", "target_statement": "Lena may register for the seminar.", "gold_reason_type": "global_closed_world_underivable", "symbolic": {"positive_atoms": ["passed_exam::Lena"], "negative_atoms": [], "rules": [{"antecedents": ["passed_exam::Lena", "has_advising_clearance::Lena"], "conclusion": "may_register::Lena", "text": "If a student passed the qualifying exam and has advising clearance, then that student may register for the seminar."}], "query_atom": "may_register::Lena", "closure_atoms": ["passed_exam::Lena"]}, "source_reports": {"agent_a": {"positive": ["Lena passed the qualifying exam."], "negative": []}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "closed_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_advising_clearance, has_external_scholarship, may_register, needs_manual_audit, passed_exam.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- has_advising_clearance: has advising clearance\n- has_external_scholarship: has an external scholarship\n- may_register: may register for the seminar\n- needs_manual_audit: needs a manual audit\n- passed_exam: passed the qualifying exam\n\nCoordinator policy rules:\n- If a student passed the qualifying exam and has advising clearance, then that student may register for the seminar.\n\nAgent A report:\nPositive facts:\n- Lena passed the qualifying exam.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: may_register::Lena.\nTarget statement: Lena may register for the seminar.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "university_registration_0056_closed_derived_missing_antecedent__lcwa__multi_agent", "source_id": "university_registration_0056_closed_derived_missing_antecedent__lcwa", "base_id": "university_registration_0056_closed_derived_missing_antecedent", "split": "dev", "domain": "university_registration", "family": "closed_derived_missing_antecedent", "subset": "core_contrastive", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_advising_clearance", "may_register"], "agent_a_complete_predicates": ["has_advising_clearance", "may_register"], "agent_b_open_predicates": ["has_external_scholarship", "needs_manual_audit", "passed_exam"], "facts_positive": ["Lena passed the qualifying exam."], "facts_negative": [], "rules_natural": ["If a student passed the qualifying exam and has advising clearance, then that student may register for the seminar."], "predicate_glossary": {"passed_exam": "passed the qualifying exam", "has_advising_clearance": "has advising clearance", "has_external_scholarship": "has an external scholarship", "may_register": "may register for the seminar", "needs_manual_audit": "needs a manual audit"}, "vocabulary_predicates": ["has_advising_clearance", "has_external_scholarship", "may_register", "needs_manual_audit", "passed_exam"], "query_atom": "may_register::Lena", "target_statement": "Lena may register for the seminar.", "gold_reason_type": "local_closed_world_underivable", "symbolic": {"positive_atoms": ["passed_exam::Lena"], "negative_atoms": [], "rules": [{"antecedents": ["passed_exam::Lena", "has_advising_clearance::Lena"], "conclusion": "may_register::Lena", "text": "If a student passed the qualifying exam and has advising clearance, then that student may register for the seminar."}], "query_atom": "may_register::Lena", "closure_atoms": ["passed_exam::Lena"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Lena passed the qualifying exam."], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "closed_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_advising_clearance, may_register.\nAgent B open/incomplete predicates: has_external_scholarship, needs_manual_audit, passed_exam.\n\nPredicate glossary:\n- has_advising_clearance: has advising clearance\n- has_external_scholarship: has an external scholarship\n- may_register: may register for the seminar\n- needs_manual_audit: needs a manual audit\n- passed_exam: passed the qualifying exam\n\nCoordinator policy rules:\n- If a student passed the qualifying exam and has advising clearance, then that student may register for the seminar.\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Lena passed the qualifying exam.\nNegative facts:\n- none\n\nTarget atom: may_register::Lena.\nTarget statement: Lena may register for the seminar.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "university_registration_0056_closed_derived_missing_antecedent__owa__multi_agent", "source_id": "university_registration_0056_closed_derived_missing_antecedent__owa", "base_id": "university_registration_0056_closed_derived_missing_antecedent", "split": "dev", "domain": "university_registration", "family": "closed_derived_missing_antecedent", "subset": "core_contrastive", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["has_advising_clearance", "has_external_scholarship", "may_register", "needs_manual_audit", "passed_exam"], "facts_positive": ["Lena passed the qualifying exam."], "facts_negative": [], "rules_natural": ["If a student passed the qualifying exam and has advising clearance, then that student may register for the seminar."], "predicate_glossary": {"passed_exam": "passed the qualifying exam", "has_advising_clearance": "has advising clearance", "has_external_scholarship": "has an external scholarship", "may_register": "may register for the seminar", "needs_manual_audit": "needs a manual audit"}, "vocabulary_predicates": ["has_advising_clearance", "has_external_scholarship", "may_register", "needs_manual_audit", "passed_exam"], "query_atom": "may_register::Lena", "target_statement": "Lena may register for the seminar.", "gold_reason_type": "open_world_underivable", "symbolic": {"positive_atoms": ["passed_exam::Lena"], "negative_atoms": [], "rules": [{"antecedents": ["passed_exam::Lena", "has_advising_clearance::Lena"], "conclusion": "may_register::Lena", "text": "If a student passed the qualifying exam and has advising clearance, then that student may register for the seminar."}], "query_atom": "may_register::Lena", "closure_atoms": ["passed_exam::Lena"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Lena passed the qualifying exam."], "negative": []}}, "gold_truth_value": "unknown", "gold_source_used": "agent_b", "gold_closure_handling": "open_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: has_advising_clearance, has_external_scholarship, may_register, needs_manual_audit, passed_exam.\n\nPredicate glossary:\n- has_advising_clearance: has advising clearance\n- has_external_scholarship: has an external scholarship\n- may_register: may register for the seminar\n- needs_manual_audit: needs a manual audit\n- passed_exam: passed the qualifying exam\n\nCoordinator policy rules:\n- If a student passed the qualifying exam and has advising clearance, then that student may register for the seminar.\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Lena passed the qualifying exam.\nNegative facts:\n- none\n\nTarget atom: may_register::Lena.\nTarget statement: Lena may register for the seminar.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "university_registration_0061_open_derived_missing_antecedent__cwa__multi_agent", "source_id": "university_registration_0061_open_derived_missing_antecedent__cwa", "base_id": "university_registration_0061_open_derived_missing_antecedent", "split": "test", "domain": "university_registration", "family": "open_derived_missing_antecedent", "subset": "core_contrastive", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["has_advising_clearance", "has_external_scholarship", "may_register", "needs_manual_audit", "passed_exam"], "agent_a_complete_predicates": ["has_advising_clearance", "has_external_scholarship", "may_register", "needs_manual_audit", "passed_exam"], "agent_b_open_predicates": [], "facts_positive": ["Noah passed the qualifying exam."], "facts_negative": [], "rules_natural": ["If a student has an external scholarship, then that student needs a manual audit."], "predicate_glossary": {"passed_exam": "passed the qualifying exam", "has_advising_clearance": "has advising clearance", "has_external_scholarship": "has an external scholarship", "may_register": "may register for the seminar", "needs_manual_audit": "needs a manual audit"}, "vocabulary_predicates": ["has_advising_clearance", "has_external_scholarship", "may_register", "needs_manual_audit", "passed_exam"], "query_atom": "needs_manual_audit::Noah", "target_statement": "Noah needs a manual audit.", "gold_reason_type": "global_closed_world_underivable", "symbolic": {"positive_atoms": ["passed_exam::Noah"], "negative_atoms": [], "rules": [{"antecedents": ["has_external_scholarship::Noah"], "conclusion": "needs_manual_audit::Noah", "text": "If a student has an external scholarship, then that student needs a manual audit."}], "query_atom": "needs_manual_audit::Noah", "closure_atoms": ["passed_exam::Noah"]}, "source_reports": {"agent_a": {"positive": ["Noah passed the qualifying exam."], "negative": []}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "closed_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_advising_clearance, has_external_scholarship, may_register, needs_manual_audit, passed_exam.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- has_advising_clearance: has advising clearance\n- has_external_scholarship: has an external scholarship\n- may_register: may register for the seminar\n- needs_manual_audit: needs a manual audit\n- passed_exam: passed the qualifying exam\n\nCoordinator policy rules:\n- If a student has an external scholarship, then that student needs a manual audit.\n\nAgent A report:\nPositive facts:\n- Noah passed the qualifying exam.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: needs_manual_audit::Noah.\nTarget statement: Noah needs a manual audit.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "university_registration_0061_open_derived_missing_antecedent__lcwa__multi_agent", "source_id": "university_registration_0061_open_derived_missing_antecedent__lcwa", "base_id": "university_registration_0061_open_derived_missing_antecedent", "split": "test", "domain": "university_registration", "family": "open_derived_missing_antecedent", "subset": "core_contrastive", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_advising_clearance", "may_register"], "agent_a_complete_predicates": ["has_advising_clearance", "may_register"], "agent_b_open_predicates": ["has_external_scholarship", "needs_manual_audit", "passed_exam"], "facts_positive": ["Noah passed the qualifying exam."], "facts_negative": [], "rules_natural": ["If a student has an external scholarship, then that student needs a manual audit."], "predicate_glossary": {"passed_exam": "passed the qualifying exam", "has_advising_clearance": "has advising clearance", "has_external_scholarship": "has an external scholarship", "may_register": "may register for the seminar", "needs_manual_audit": "needs a manual audit"}, "vocabulary_predicates": ["has_advising_clearance", "has_external_scholarship", "may_register", "needs_manual_audit", "passed_exam"], "query_atom": "needs_manual_audit::Noah", "target_statement": "Noah needs a manual audit.", "gold_reason_type": "local_open_world_underivable", "symbolic": {"positive_atoms": ["passed_exam::Noah"], "negative_atoms": [], "rules": [{"antecedents": ["has_external_scholarship::Noah"], "conclusion": "needs_manual_audit::Noah", "text": "If a student has an external scholarship, then that student needs a manual audit."}], "query_atom": "needs_manual_audit::Noah", "closure_atoms": ["passed_exam::Noah"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Noah passed the qualifying exam."], "negative": []}}, "gold_truth_value": "unknown", "gold_source_used": "agent_b", "gold_closure_handling": "open_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_advising_clearance, may_register.\nAgent B open/incomplete predicates: has_external_scholarship, needs_manual_audit, passed_exam.\n\nPredicate glossary:\n- has_advising_clearance: has advising clearance\n- has_external_scholarship: has an external scholarship\n- may_register: may register for the seminar\n- needs_manual_audit: needs a manual audit\n- passed_exam: passed the qualifying exam\n\nCoordinator policy rules:\n- If a student has an external scholarship, then that student needs a manual audit.\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Noah passed the qualifying exam.\nNegative facts:\n- none\n\nTarget atom: needs_manual_audit::Noah.\nTarget statement: Noah needs a manual audit.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "university_registration_0061_open_derived_missing_antecedent__owa__multi_agent", "source_id": "university_registration_0061_open_derived_missing_antecedent__owa", "base_id": "university_registration_0061_open_derived_missing_antecedent", "split": "test", "domain": "university_registration", "family": "open_derived_missing_antecedent", "subset": "core_contrastive", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["has_advising_clearance", "has_external_scholarship", "may_register", "needs_manual_audit", "passed_exam"], "facts_positive": ["Noah passed the qualifying exam."], "facts_negative": [], "rules_natural": ["If a student has an external scholarship, then that student needs a manual audit."], "predicate_glossary": {"passed_exam": "passed the qualifying exam", "has_advising_clearance": "has advising clearance", "has_external_scholarship": "has an external scholarship", "may_register": "may register for the seminar", "needs_manual_audit": "needs a manual audit"}, "vocabulary_predicates": ["has_advising_clearance", "has_external_scholarship", "may_register", "needs_manual_audit", "passed_exam"], "query_atom": "needs_manual_audit::Noah", "target_statement": "Noah needs a manual audit.", "gold_reason_type": "open_world_underivable", "symbolic": {"positive_atoms": ["passed_exam::Noah"], "negative_atoms": [], "rules": [{"antecedents": ["has_external_scholarship::Noah"], "conclusion": "needs_manual_audit::Noah", "text": "If a student has an external scholarship, then that student needs a manual audit."}], "query_atom": "needs_manual_audit::Noah", "closure_atoms": ["passed_exam::Noah"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Noah passed the qualifying exam."], "negative": []}}, "gold_truth_value": "unknown", "gold_source_used": "agent_b", "gold_closure_handling": "open_absence", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: has_advising_clearance, has_external_scholarship, may_register, needs_manual_audit, passed_exam.\n\nPredicate glossary:\n- has_advising_clearance: has advising clearance\n- has_external_scholarship: has an external scholarship\n- may_register: may register for the seminar\n- needs_manual_audit: needs a manual audit\n- passed_exam: passed the qualifying exam\n\nCoordinator policy rules:\n- If a student has an external scholarship, then that student needs a manual audit.\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Noah passed the qualifying exam.\nNegative facts:\n- none\n\nTarget atom: needs_manual_audit::Noah.\nTarget statement: Noah needs a manual audit.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "university_registration_0064_entailed_closed_conclusion__cwa__multi_agent", "source_id": "university_registration_0064_entailed_closed_conclusion__cwa", "base_id": "university_registration_0064_entailed_closed_conclusion", "split": "train", "domain": "university_registration", "family": "entailed_closed_conclusion", "subset": "control_entailed", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["has_advising_clearance", "has_external_scholarship", "may_register", "needs_manual_audit", "passed_exam"], "agent_a_complete_predicates": ["has_advising_clearance", "has_external_scholarship", "may_register", "needs_manual_audit", "passed_exam"], "agent_b_open_predicates": [], "facts_positive": ["Iris has advising clearance.", "Iris passed the qualifying exam."], "facts_negative": [], "rules_natural": ["If a student passed the qualifying exam and has advising clearance, then that student may register for the seminar."], "predicate_glossary": {"passed_exam": "passed the qualifying exam", "has_advising_clearance": "has advising clearance", "has_external_scholarship": "has an external scholarship", "may_register": "may register for the seminar", "needs_manual_audit": "needs a manual audit"}, "vocabulary_predicates": ["has_advising_clearance", "has_external_scholarship", "may_register", "needs_manual_audit", "passed_exam"], "query_atom": "may_register::Iris", "target_statement": "Iris may register for the seminar.", "gold_reason_type": "entailed_by_fact_or_rule", "symbolic": {"positive_atoms": ["has_advising_clearance::Iris", "passed_exam::Iris"], "negative_atoms": [], "rules": [{"antecedents": ["passed_exam::Iris", "has_advising_clearance::Iris"], "conclusion": "may_register::Iris", "text": "If a student passed the qualifying exam and has advising clearance, then that student may register for the seminar."}], "query_atom": "may_register::Iris", "closure_atoms": ["has_advising_clearance::Iris", "may_register::Iris", "passed_exam::Iris"]}, "source_reports": {"agent_a": {"positive": ["Iris has advising clearance.", "Iris passed the qualifying exam."], "negative": []}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "true", "gold_source_used": "agent_a", "gold_closure_handling": "entailed_or_explicit", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_advising_clearance, has_external_scholarship, may_register, needs_manual_audit, passed_exam.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- has_advising_clearance: has advising clearance\n- has_external_scholarship: has an external scholarship\n- may_register: may register for the seminar\n- needs_manual_audit: needs a manual audit\n- passed_exam: passed the qualifying exam\n\nCoordinator policy rules:\n- If a student passed the qualifying exam and has advising clearance, then that student may register for the seminar.\n\nAgent A report:\nPositive facts:\n- Iris has advising clearance.\n- Iris passed the qualifying exam.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: may_register::Iris.\nTarget statement: Iris may register for the seminar.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "university_registration_0064_entailed_closed_conclusion__lcwa__multi_agent", "source_id": "university_registration_0064_entailed_closed_conclusion__lcwa", "base_id": "university_registration_0064_entailed_closed_conclusion", "split": "train", "domain": "university_registration", "family": "entailed_closed_conclusion", "subset": "control_entailed", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_advising_clearance", "may_register"], "agent_a_complete_predicates": ["has_advising_clearance", "may_register"], "agent_b_open_predicates": ["has_external_scholarship", "needs_manual_audit", "passed_exam"], "facts_positive": ["Iris has advising clearance.", "Iris passed the qualifying exam."], "facts_negative": [], "rules_natural": ["If a student passed the qualifying exam and has advising clearance, then that student may register for the seminar."], "predicate_glossary": {"passed_exam": "passed the qualifying exam", "has_advising_clearance": "has advising clearance", "has_external_scholarship": "has an external scholarship", "may_register": "may register for the seminar", "needs_manual_audit": "needs a manual audit"}, "vocabulary_predicates": ["has_advising_clearance", "has_external_scholarship", "may_register", "needs_manual_audit", "passed_exam"], "query_atom": "may_register::Iris", "target_statement": "Iris may register for the seminar.", "gold_reason_type": "entailed_by_fact_or_rule", "symbolic": {"positive_atoms": ["has_advising_clearance::Iris", "passed_exam::Iris"], "negative_atoms": [], "rules": [{"antecedents": ["passed_exam::Iris", "has_advising_clearance::Iris"], "conclusion": "may_register::Iris", "text": "If a student passed the qualifying exam and has advising clearance, then that student may register for the seminar."}], "query_atom": "may_register::Iris", "closure_atoms": ["has_advising_clearance::Iris", "may_register::Iris", "passed_exam::Iris"]}, "source_reports": {"agent_a": {"positive": ["Iris has advising clearance."], "negative": []}, "agent_b": {"positive": ["Iris passed the qualifying exam."], "negative": []}}, "gold_truth_value": "true", "gold_source_used": "both", "gold_closure_handling": "entailed_or_explicit", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_advising_clearance, may_register.\nAgent B open/incomplete predicates: has_external_scholarship, needs_manual_audit, passed_exam.\n\nPredicate glossary:\n- has_advising_clearance: has advising clearance\n- has_external_scholarship: has an external scholarship\n- may_register: may register for the seminar\n- needs_manual_audit: needs a manual audit\n- passed_exam: passed the qualifying exam\n\nCoordinator policy rules:\n- If a student passed the qualifying exam and has advising clearance, then that student may register for the seminar.\n\nAgent A report:\nPositive facts:\n- Iris has advising clearance.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Iris passed the qualifying exam.\nNegative facts:\n- none\n\nTarget atom: may_register::Iris.\nTarget statement: Iris may register for the seminar.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "university_registration_0064_entailed_closed_conclusion__owa__multi_agent", "source_id": "university_registration_0064_entailed_closed_conclusion__owa", "base_id": "university_registration_0064_entailed_closed_conclusion", "split": "train", "domain": "university_registration", "family": "entailed_closed_conclusion", "subset": "control_entailed", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["has_advising_clearance", "has_external_scholarship", "may_register", "needs_manual_audit", "passed_exam"], "facts_positive": ["Iris has advising clearance.", "Iris passed the qualifying exam."], "facts_negative": [], "rules_natural": ["If a student passed the qualifying exam and has advising clearance, then that student may register for the seminar."], "predicate_glossary": {"passed_exam": "passed the qualifying exam", "has_advising_clearance": "has advising clearance", "has_external_scholarship": "has an external scholarship", "may_register": "may register for the seminar", "needs_manual_audit": "needs a manual audit"}, "vocabulary_predicates": ["has_advising_clearance", "has_external_scholarship", "may_register", "needs_manual_audit", "passed_exam"], "query_atom": "may_register::Iris", "target_statement": "Iris may register for the seminar.", "gold_reason_type": "entailed_by_fact_or_rule", "symbolic": {"positive_atoms": ["has_advising_clearance::Iris", "passed_exam::Iris"], "negative_atoms": [], "rules": [{"antecedents": ["passed_exam::Iris", "has_advising_clearance::Iris"], "conclusion": "may_register::Iris", "text": "If a student passed the qualifying exam and has advising clearance, then that student may register for the seminar."}], "query_atom": "may_register::Iris", "closure_atoms": ["has_advising_clearance::Iris", "may_register::Iris", "passed_exam::Iris"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Iris has advising clearance.", "Iris passed the qualifying exam."], "negative": []}}, "gold_truth_value": "true", "gold_source_used": "agent_b", "gold_closure_handling": "entailed_or_explicit", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: has_advising_clearance, has_external_scholarship, may_register, needs_manual_audit, passed_exam.\n\nPredicate glossary:\n- has_advising_clearance: has advising clearance\n- has_external_scholarship: has an external scholarship\n- may_register: may register for the seminar\n- needs_manual_audit: needs a manual audit\n- passed_exam: passed the qualifying exam\n\nCoordinator policy rules:\n- If a student passed the qualifying exam and has advising clearance, then that student may register for the seminar.\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Iris has advising clearance.\n- Iris passed the qualifying exam.\nNegative facts:\n- none\n\nTarget atom: may_register::Iris.\nTarget statement: Iris may register for the seminar.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "university_registration_0065_entailed_closed_conclusion__cwa__multi_agent", "source_id": "university_registration_0065_entailed_closed_conclusion__cwa", "base_id": "university_registration_0065_entailed_closed_conclusion", "split": "test", "domain": "university_registration", "family": "entailed_closed_conclusion", "subset": "control_entailed", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["has_advising_clearance", "has_external_scholarship", "may_register", "needs_manual_audit", "passed_exam"], "agent_a_complete_predicates": ["has_advising_clearance", "has_external_scholarship", "may_register", "needs_manual_audit", "passed_exam"], "agent_b_open_predicates": [], "facts_positive": ["Theo has advising clearance.", "Theo passed the qualifying exam."], "facts_negative": [], "rules_natural": ["If a student passed the qualifying exam and has advising clearance, then that student may register for the seminar."], "predicate_glossary": {"passed_exam": "passed the qualifying exam", "has_advising_clearance": "has advising clearance", "has_external_scholarship": "has an external scholarship", "may_register": "may register for the seminar", "needs_manual_audit": "needs a manual audit"}, "vocabulary_predicates": ["has_advising_clearance", "has_external_scholarship", "may_register", "needs_manual_audit", "passed_exam"], "query_atom": "may_register::Theo", "target_statement": "Theo may register for the seminar.", "gold_reason_type": "entailed_by_fact_or_rule", "symbolic": {"positive_atoms": ["has_advising_clearance::Theo", "passed_exam::Theo"], "negative_atoms": [], "rules": [{"antecedents": ["passed_exam::Theo", "has_advising_clearance::Theo"], "conclusion": "may_register::Theo", "text": "If a student passed the qualifying exam and has advising clearance, then that student may register for the seminar."}], "query_atom": "may_register::Theo", "closure_atoms": ["has_advising_clearance::Theo", "may_register::Theo", "passed_exam::Theo"]}, "source_reports": {"agent_a": {"positive": ["Theo has advising clearance.", "Theo passed the qualifying exam."], "negative": []}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "true", "gold_source_used": "agent_a", "gold_closure_handling": "entailed_or_explicit", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_advising_clearance, has_external_scholarship, may_register, needs_manual_audit, passed_exam.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- has_advising_clearance: has advising clearance\n- has_external_scholarship: has an external scholarship\n- may_register: may register for the seminar\n- needs_manual_audit: needs a manual audit\n- passed_exam: passed the qualifying exam\n\nCoordinator policy rules:\n- If a student passed the qualifying exam and has advising clearance, then that student may register for the seminar.\n\nAgent A report:\nPositive facts:\n- Theo has advising clearance.\n- Theo passed the qualifying exam.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: may_register::Theo.\nTarget statement: Theo may register for the seminar.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "university_registration_0065_entailed_closed_conclusion__lcwa__multi_agent", "source_id": "university_registration_0065_entailed_closed_conclusion__lcwa", "base_id": "university_registration_0065_entailed_closed_conclusion", "split": "test", "domain": "university_registration", "family": "entailed_closed_conclusion", "subset": "control_entailed", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_advising_clearance", "may_register"], "agent_a_complete_predicates": ["has_advising_clearance", "may_register"], "agent_b_open_predicates": ["has_external_scholarship", "needs_manual_audit", "passed_exam"], "facts_positive": ["Theo has advising clearance.", "Theo passed the qualifying exam."], "facts_negative": [], "rules_natural": ["If a student passed the qualifying exam and has advising clearance, then that student may register for the seminar."], "predicate_glossary": {"passed_exam": "passed the qualifying exam", "has_advising_clearance": "has advising clearance", "has_external_scholarship": "has an external scholarship", "may_register": "may register for the seminar", "needs_manual_audit": "needs a manual audit"}, "vocabulary_predicates": ["has_advising_clearance", "has_external_scholarship", "may_register", "needs_manual_audit", "passed_exam"], "query_atom": "may_register::Theo", "target_statement": "Theo may register for the seminar.", "gold_reason_type": "entailed_by_fact_or_rule", "symbolic": {"positive_atoms": ["has_advising_clearance::Theo", "passed_exam::Theo"], "negative_atoms": [], "rules": [{"antecedents": ["passed_exam::Theo", "has_advising_clearance::Theo"], "conclusion": "may_register::Theo", "text": "If a student passed the qualifying exam and has advising clearance, then that student may register for the seminar."}], "query_atom": "may_register::Theo", "closure_atoms": ["has_advising_clearance::Theo", "may_register::Theo", "passed_exam::Theo"]}, "source_reports": {"agent_a": {"positive": ["Theo has advising clearance."], "negative": []}, "agent_b": {"positive": ["Theo passed the qualifying exam."], "negative": []}}, "gold_truth_value": "true", "gold_source_used": "both", "gold_closure_handling": "entailed_or_explicit", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_advising_clearance, may_register.\nAgent B open/incomplete predicates: has_external_scholarship, needs_manual_audit, passed_exam.\n\nPredicate glossary:\n- has_advising_clearance: has advising clearance\n- has_external_scholarship: has an external scholarship\n- may_register: may register for the seminar\n- needs_manual_audit: needs a manual audit\n- passed_exam: passed the qualifying exam\n\nCoordinator policy rules:\n- If a student passed the qualifying exam and has advising clearance, then that student may register for the seminar.\n\nAgent A report:\nPositive facts:\n- Theo has advising clearance.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Theo passed the qualifying exam.\nNegative facts:\n- none\n\nTarget atom: may_register::Theo.\nTarget statement: Theo may register for the seminar.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "university_registration_0065_entailed_closed_conclusion__owa__multi_agent", "source_id": "university_registration_0065_entailed_closed_conclusion__owa", "base_id": "university_registration_0065_entailed_closed_conclusion", "split": "test", "domain": "university_registration", "family": "entailed_closed_conclusion", "subset": "control_entailed", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["has_advising_clearance", "has_external_scholarship", "may_register", "needs_manual_audit", "passed_exam"], "facts_positive": ["Theo has advising clearance.", "Theo passed the qualifying exam."], "facts_negative": [], "rules_natural": ["If a student passed the qualifying exam and has advising clearance, then that student may register for the seminar."], "predicate_glossary": {"passed_exam": "passed the qualifying exam", "has_advising_clearance": "has advising clearance", "has_external_scholarship": "has an external scholarship", "may_register": "may register for the seminar", "needs_manual_audit": "needs a manual audit"}, "vocabulary_predicates": ["has_advising_clearance", "has_external_scholarship", "may_register", "needs_manual_audit", "passed_exam"], "query_atom": "may_register::Theo", "target_statement": "Theo may register for the seminar.", "gold_reason_type": "entailed_by_fact_or_rule", "symbolic": {"positive_atoms": ["has_advising_clearance::Theo", "passed_exam::Theo"], "negative_atoms": [], "rules": [{"antecedents": ["passed_exam::Theo", "has_advising_clearance::Theo"], "conclusion": "may_register::Theo", "text": "If a student passed the qualifying exam and has advising clearance, then that student may register for the seminar."}], "query_atom": "may_register::Theo", "closure_atoms": ["has_advising_clearance::Theo", "may_register::Theo", "passed_exam::Theo"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Theo has advising clearance.", "Theo passed the qualifying exam."], "negative": []}}, "gold_truth_value": "true", "gold_source_used": "agent_b", "gold_closure_handling": "entailed_or_explicit", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: has_advising_clearance, has_external_scholarship, may_register, needs_manual_audit, passed_exam.\n\nPredicate glossary:\n- has_advising_clearance: has advising clearance\n- has_external_scholarship: has an external scholarship\n- may_register: may register for the seminar\n- needs_manual_audit: needs a manual audit\n- passed_exam: passed the qualifying exam\n\nCoordinator policy rules:\n- If a student passed the qualifying exam and has advising clearance, then that student may register for the seminar.\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Theo has advising clearance.\n- Theo passed the qualifying exam.\nNegative facts:\n- none\n\nTarget atom: may_register::Theo.\nTarget statement: Theo may register for the seminar.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "university_registration_0066_entailed_closed_conclusion__cwa__multi_agent", "source_id": "university_registration_0066_entailed_closed_conclusion__cwa", "base_id": "university_registration_0066_entailed_closed_conclusion", "split": "train", "domain": "university_registration", "family": "entailed_closed_conclusion", "subset": "control_entailed", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["has_advising_clearance", "has_external_scholarship", "may_register", "needs_manual_audit", "passed_exam"], "agent_a_complete_predicates": ["has_advising_clearance", "has_external_scholarship", "may_register", "needs_manual_audit", "passed_exam"], "agent_b_open_predicates": [], "facts_positive": ["Mia has advising clearance.", "Mia passed the qualifying exam."], "facts_negative": [], "rules_natural": ["If a student passed the qualifying exam and has advising clearance, then that student may register for the seminar."], "predicate_glossary": {"passed_exam": "passed the qualifying exam", "has_advising_clearance": "has advising clearance", "has_external_scholarship": "has an external scholarship", "may_register": "may register for the seminar", "needs_manual_audit": "needs a manual audit"}, "vocabulary_predicates": ["has_advising_clearance", "has_external_scholarship", "may_register", "needs_manual_audit", "passed_exam"], "query_atom": "may_register::Mia", "target_statement": "Mia may register for the seminar.", "gold_reason_type": "entailed_by_fact_or_rule", "symbolic": {"positive_atoms": ["has_advising_clearance::Mia", "passed_exam::Mia"], "negative_atoms": [], "rules": [{"antecedents": ["passed_exam::Mia", "has_advising_clearance::Mia"], "conclusion": "may_register::Mia", "text": "If a student passed the qualifying exam and has advising clearance, then that student may register for the seminar."}], "query_atom": "may_register::Mia", "closure_atoms": ["has_advising_clearance::Mia", "may_register::Mia", "passed_exam::Mia"]}, "source_reports": {"agent_a": {"positive": ["Mia has advising clearance.", "Mia passed the qualifying exam."], "negative": []}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "true", "gold_source_used": "agent_a", "gold_closure_handling": "entailed_or_explicit", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_advising_clearance, has_external_scholarship, may_register, needs_manual_audit, passed_exam.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- has_advising_clearance: has advising clearance\n- has_external_scholarship: has an external scholarship\n- may_register: may register for the seminar\n- needs_manual_audit: needs a manual audit\n- passed_exam: passed the qualifying exam\n\nCoordinator policy rules:\n- If a student passed the qualifying exam and has advising clearance, then that student may register for the seminar.\n\nAgent A report:\nPositive facts:\n- Mia has advising clearance.\n- Mia passed the qualifying exam.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: may_register::Mia.\nTarget statement: Mia may register for the seminar.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "university_registration_0066_entailed_closed_conclusion__lcwa__multi_agent", "source_id": "university_registration_0066_entailed_closed_conclusion__lcwa", "base_id": "university_registration_0066_entailed_closed_conclusion", "split": "train", "domain": "university_registration", "family": "entailed_closed_conclusion", "subset": "control_entailed", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_advising_clearance", "may_register"], "agent_a_complete_predicates": ["has_advising_clearance", "may_register"], "agent_b_open_predicates": ["has_external_scholarship", "needs_manual_audit", "passed_exam"], "facts_positive": ["Mia has advising clearance.", "Mia passed the qualifying exam."], "facts_negative": [], "rules_natural": ["If a student passed the qualifying exam and has advising clearance, then that student may register for the seminar."], "predicate_glossary": {"passed_exam": "passed the qualifying exam", "has_advising_clearance": "has advising clearance", "has_external_scholarship": "has an external scholarship", "may_register": "may register for the seminar", "needs_manual_audit": "needs a manual audit"}, "vocabulary_predicates": ["has_advising_clearance", "has_external_scholarship", "may_register", "needs_manual_audit", "passed_exam"], "query_atom": "may_register::Mia", "target_statement": "Mia may register for the seminar.", "gold_reason_type": "entailed_by_fact_or_rule", "symbolic": {"positive_atoms": ["has_advising_clearance::Mia", "passed_exam::Mia"], "negative_atoms": [], "rules": [{"antecedents": ["passed_exam::Mia", "has_advising_clearance::Mia"], "conclusion": "may_register::Mia", "text": "If a student passed the qualifying exam and has advising clearance, then that student may register for the seminar."}], "query_atom": "may_register::Mia", "closure_atoms": ["has_advising_clearance::Mia", "may_register::Mia", "passed_exam::Mia"]}, "source_reports": {"agent_a": {"positive": ["Mia has advising clearance."], "negative": []}, "agent_b": {"positive": ["Mia passed the qualifying exam."], "negative": []}}, "gold_truth_value": "true", "gold_source_used": "both", "gold_closure_handling": "entailed_or_explicit", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_advising_clearance, may_register.\nAgent B open/incomplete predicates: has_external_scholarship, needs_manual_audit, passed_exam.\n\nPredicate glossary:\n- has_advising_clearance: has advising clearance\n- has_external_scholarship: has an external scholarship\n- may_register: may register for the seminar\n- needs_manual_audit: needs a manual audit\n- passed_exam: passed the qualifying exam\n\nCoordinator policy rules:\n- If a student passed the qualifying exam and has advising clearance, then that student may register for the seminar.\n\nAgent A report:\nPositive facts:\n- Mia has advising clearance.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Mia passed the qualifying exam.\nNegative facts:\n- none\n\nTarget atom: may_register::Mia.\nTarget statement: Mia may register for the seminar.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "university_registration_0066_entailed_closed_conclusion__owa__multi_agent", "source_id": "university_registration_0066_entailed_closed_conclusion__owa", "base_id": "university_registration_0066_entailed_closed_conclusion", "split": "train", "domain": "university_registration", "family": "entailed_closed_conclusion", "subset": "control_entailed", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["has_advising_clearance", "has_external_scholarship", "may_register", "needs_manual_audit", "passed_exam"], "facts_positive": ["Mia has advising clearance.", "Mia passed the qualifying exam."], "facts_negative": [], "rules_natural": ["If a student passed the qualifying exam and has advising clearance, then that student may register for the seminar."], "predicate_glossary": {"passed_exam": "passed the qualifying exam", "has_advising_clearance": "has advising clearance", "has_external_scholarship": "has an external scholarship", "may_register": "may register for the seminar", "needs_manual_audit": "needs a manual audit"}, "vocabulary_predicates": ["has_advising_clearance", "has_external_scholarship", "may_register", "needs_manual_audit", "passed_exam"], "query_atom": "may_register::Mia", "target_statement": "Mia may register for the seminar.", "gold_reason_type": "entailed_by_fact_or_rule", "symbolic": {"positive_atoms": ["has_advising_clearance::Mia", "passed_exam::Mia"], "negative_atoms": [], "rules": [{"antecedents": ["passed_exam::Mia", "has_advising_clearance::Mia"], "conclusion": "may_register::Mia", "text": "If a student passed the qualifying exam and has advising clearance, then that student may register for the seminar."}], "query_atom": "may_register::Mia", "closure_atoms": ["has_advising_clearance::Mia", "may_register::Mia", "passed_exam::Mia"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Mia has advising clearance.", "Mia passed the qualifying exam."], "negative": []}}, "gold_truth_value": "true", "gold_source_used": "agent_b", "gold_closure_handling": "entailed_or_explicit", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: has_advising_clearance, has_external_scholarship, may_register, needs_manual_audit, passed_exam.\n\nPredicate glossary:\n- has_advising_clearance: has advising clearance\n- has_external_scholarship: has an external scholarship\n- may_register: may register for the seminar\n- needs_manual_audit: needs a manual audit\n- passed_exam: passed the qualifying exam\n\nCoordinator policy rules:\n- If a student passed the qualifying exam and has advising clearance, then that student may register for the seminar.\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Mia has advising clearance.\n- Mia passed the qualifying exam.\nNegative facts:\n- none\n\nTarget atom: may_register::Mia.\nTarget statement: Mia may register for the seminar.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "university_registration_0067_entailed_closed_conclusion__cwa__multi_agent", "source_id": "university_registration_0067_entailed_closed_conclusion__cwa", "base_id": "university_registration_0067_entailed_closed_conclusion", "split": "test", "domain": "university_registration", "family": "entailed_closed_conclusion", "subset": "control_entailed", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["has_advising_clearance", "has_external_scholarship", "may_register", "needs_manual_audit", "passed_exam"], "agent_a_complete_predicates": ["has_advising_clearance", "has_external_scholarship", "may_register", "needs_manual_audit", "passed_exam"], "agent_b_open_predicates": [], "facts_positive": ["Noah has advising clearance.", "Noah passed the qualifying exam."], "facts_negative": [], "rules_natural": ["If a student passed the qualifying exam and has advising clearance, then that student may register for the seminar."], "predicate_glossary": {"passed_exam": "passed the qualifying exam", "has_advising_clearance": "has advising clearance", "has_external_scholarship": "has an external scholarship", "may_register": "may register for the seminar", "needs_manual_audit": "needs a manual audit"}, "vocabulary_predicates": ["has_advising_clearance", "has_external_scholarship", "may_register", "needs_manual_audit", "passed_exam"], "query_atom": "may_register::Noah", "target_statement": "Noah may register for the seminar.", "gold_reason_type": "entailed_by_fact_or_rule", "symbolic": {"positive_atoms": ["has_advising_clearance::Noah", "passed_exam::Noah"], "negative_atoms": [], "rules": [{"antecedents": ["passed_exam::Noah", "has_advising_clearance::Noah"], "conclusion": "may_register::Noah", "text": "If a student passed the qualifying exam and has advising clearance, then that student may register for the seminar."}], "query_atom": "may_register::Noah", "closure_atoms": ["has_advising_clearance::Noah", "may_register::Noah", "passed_exam::Noah"]}, "source_reports": {"agent_a": {"positive": ["Noah has advising clearance.", "Noah passed the qualifying exam."], "negative": []}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "true", "gold_source_used": "agent_a", "gold_closure_handling": "entailed_or_explicit", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_advising_clearance, has_external_scholarship, may_register, needs_manual_audit, passed_exam.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- has_advising_clearance: has advising clearance\n- has_external_scholarship: has an external scholarship\n- may_register: may register for the seminar\n- needs_manual_audit: needs a manual audit\n- passed_exam: passed the qualifying exam\n\nCoordinator policy rules:\n- If a student passed the qualifying exam and has advising clearance, then that student may register for the seminar.\n\nAgent A report:\nPositive facts:\n- Noah has advising clearance.\n- Noah passed the qualifying exam.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: may_register::Noah.\nTarget statement: Noah may register for the seminar.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "university_registration_0067_entailed_closed_conclusion__lcwa__multi_agent", "source_id": "university_registration_0067_entailed_closed_conclusion__lcwa", "base_id": "university_registration_0067_entailed_closed_conclusion", "split": "test", "domain": "university_registration", "family": "entailed_closed_conclusion", "subset": "control_entailed", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_advising_clearance", "may_register"], "agent_a_complete_predicates": ["has_advising_clearance", "may_register"], "agent_b_open_predicates": ["has_external_scholarship", "needs_manual_audit", "passed_exam"], "facts_positive": ["Noah has advising clearance.", "Noah passed the qualifying exam."], "facts_negative": [], "rules_natural": ["If a student passed the qualifying exam and has advising clearance, then that student may register for the seminar."], "predicate_glossary": {"passed_exam": "passed the qualifying exam", "has_advising_clearance": "has advising clearance", "has_external_scholarship": "has an external scholarship", "may_register": "may register for the seminar", "needs_manual_audit": "needs a manual audit"}, "vocabulary_predicates": ["has_advising_clearance", "has_external_scholarship", "may_register", "needs_manual_audit", "passed_exam"], "query_atom": "may_register::Noah", "target_statement": "Noah may register for the seminar.", "gold_reason_type": "entailed_by_fact_or_rule", "symbolic": {"positive_atoms": ["has_advising_clearance::Noah", "passed_exam::Noah"], "negative_atoms": [], "rules": [{"antecedents": ["passed_exam::Noah", "has_advising_clearance::Noah"], "conclusion": "may_register::Noah", "text": "If a student passed the qualifying exam and has advising clearance, then that student may register for the seminar."}], "query_atom": "may_register::Noah", "closure_atoms": ["has_advising_clearance::Noah", "may_register::Noah", "passed_exam::Noah"]}, "source_reports": {"agent_a": {"positive": ["Noah has advising clearance."], "negative": []}, "agent_b": {"positive": ["Noah passed the qualifying exam."], "negative": []}}, "gold_truth_value": "true", "gold_source_used": "both", "gold_closure_handling": "entailed_or_explicit", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_advising_clearance, may_register.\nAgent B open/incomplete predicates: has_external_scholarship, needs_manual_audit, passed_exam.\n\nPredicate glossary:\n- has_advising_clearance: has advising clearance\n- has_external_scholarship: has an external scholarship\n- may_register: may register for the seminar\n- needs_manual_audit: needs a manual audit\n- passed_exam: passed the qualifying exam\n\nCoordinator policy rules:\n- If a student passed the qualifying exam and has advising clearance, then that student may register for the seminar.\n\nAgent A report:\nPositive facts:\n- Noah has advising clearance.\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Noah passed the qualifying exam.\nNegative facts:\n- none\n\nTarget atom: may_register::Noah.\nTarget statement: Noah may register for the seminar.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "university_registration_0067_entailed_closed_conclusion__owa__multi_agent", "source_id": "university_registration_0067_entailed_closed_conclusion__owa", "base_id": "university_registration_0067_entailed_closed_conclusion", "split": "test", "domain": "university_registration", "family": "entailed_closed_conclusion", "subset": "control_entailed", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["has_advising_clearance", "has_external_scholarship", "may_register", "needs_manual_audit", "passed_exam"], "facts_positive": ["Noah has advising clearance.", "Noah passed the qualifying exam."], "facts_negative": [], "rules_natural": ["If a student passed the qualifying exam and has advising clearance, then that student may register for the seminar."], "predicate_glossary": {"passed_exam": "passed the qualifying exam", "has_advising_clearance": "has advising clearance", "has_external_scholarship": "has an external scholarship", "may_register": "may register for the seminar", "needs_manual_audit": "needs a manual audit"}, "vocabulary_predicates": ["has_advising_clearance", "has_external_scholarship", "may_register", "needs_manual_audit", "passed_exam"], "query_atom": "may_register::Noah", "target_statement": "Noah may register for the seminar.", "gold_reason_type": "entailed_by_fact_or_rule", "symbolic": {"positive_atoms": ["has_advising_clearance::Noah", "passed_exam::Noah"], "negative_atoms": [], "rules": [{"antecedents": ["passed_exam::Noah", "has_advising_clearance::Noah"], "conclusion": "may_register::Noah", "text": "If a student passed the qualifying exam and has advising clearance, then that student may register for the seminar."}], "query_atom": "may_register::Noah", "closure_atoms": ["has_advising_clearance::Noah", "may_register::Noah", "passed_exam::Noah"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Noah has advising clearance.", "Noah passed the qualifying exam."], "negative": []}}, "gold_truth_value": "true", "gold_source_used": "agent_b", "gold_closure_handling": "entailed_or_explicit", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: has_advising_clearance, has_external_scholarship, may_register, needs_manual_audit, passed_exam.\n\nPredicate glossary:\n- has_advising_clearance: has advising clearance\n- has_external_scholarship: has an external scholarship\n- may_register: may register for the seminar\n- needs_manual_audit: needs a manual audit\n- passed_exam: passed the qualifying exam\n\nCoordinator policy rules:\n- If a student passed the qualifying exam and has advising clearance, then that student may register for the seminar.\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Noah has advising clearance.\n- Noah passed the qualifying exam.\nNegative facts:\n- none\n\nTarget atom: may_register::Noah.\nTarget statement: Noah may register for the seminar.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "university_registration_0072_explicit_negative_closed__cwa__multi_agent", "source_id": "university_registration_0072_explicit_negative_closed__cwa", "base_id": "university_registration_0072_explicit_negative_closed", "split": "test", "domain": "university_registration", "family": "explicit_negative_closed", "subset": "control_explicit_negative", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["has_advising_clearance", "has_external_scholarship", "may_register", "needs_manual_audit", "passed_exam"], "agent_a_complete_predicates": ["has_advising_clearance", "has_external_scholarship", "may_register", "needs_manual_audit", "passed_exam"], "agent_b_open_predicates": [], "facts_positive": ["Lena passed the qualifying exam."], "facts_negative": ["Mia does not has advising clearance."], "rules_natural": [], "predicate_glossary": {"passed_exam": "passed the qualifying exam", "has_advising_clearance": "has advising clearance", "has_external_scholarship": "has an external scholarship", "may_register": "may register for the seminar", "needs_manual_audit": "needs a manual audit"}, "vocabulary_predicates": ["has_advising_clearance", "has_external_scholarship", "may_register", "needs_manual_audit", "passed_exam"], "query_atom": "has_advising_clearance::Mia", "target_statement": "Mia has advising clearance.", "gold_reason_type": "explicit_negative_fact", "symbolic": {"positive_atoms": ["passed_exam::Lena"], "negative_atoms": ["has_advising_clearance::Mia"], "rules": [], "query_atom": "has_advising_clearance::Mia", "closure_atoms": ["passed_exam::Lena"]}, "source_reports": {"agent_a": {"positive": ["Lena passed the qualifying exam."], "negative": ["Mia does not has advising clearance."]}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "explicit_negative", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_advising_clearance, has_external_scholarship, may_register, needs_manual_audit, passed_exam.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- has_advising_clearance: has advising clearance\n- has_external_scholarship: has an external scholarship\n- may_register: may register for the seminar\n- needs_manual_audit: needs a manual audit\n- passed_exam: passed the qualifying exam\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- Lena passed the qualifying exam.\nNegative facts:\n- Mia does not has advising clearance.\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: has_advising_clearance::Mia.\nTarget statement: Mia has advising clearance.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "university_registration_0072_explicit_negative_closed__lcwa__multi_agent", "source_id": "university_registration_0072_explicit_negative_closed__lcwa", "base_id": "university_registration_0072_explicit_negative_closed", "split": "test", "domain": "university_registration", "family": "explicit_negative_closed", "subset": "control_explicit_negative", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_advising_clearance"], "agent_a_complete_predicates": ["has_advising_clearance"], "agent_b_open_predicates": ["has_external_scholarship", "may_register", "needs_manual_audit", "passed_exam"], "facts_positive": ["Lena passed the qualifying exam."], "facts_negative": ["Mia does not has advising clearance."], "rules_natural": [], "predicate_glossary": {"passed_exam": "passed the qualifying exam", "has_advising_clearance": "has advising clearance", "has_external_scholarship": "has an external scholarship", "may_register": "may register for the seminar", "needs_manual_audit": "needs a manual audit"}, "vocabulary_predicates": ["has_advising_clearance", "has_external_scholarship", "may_register", "needs_manual_audit", "passed_exam"], "query_atom": "has_advising_clearance::Mia", "target_statement": "Mia has advising clearance.", "gold_reason_type": "explicit_negative_fact", "symbolic": {"positive_atoms": ["passed_exam::Lena"], "negative_atoms": ["has_advising_clearance::Mia"], "rules": [], "query_atom": "has_advising_clearance::Mia", "closure_atoms": ["passed_exam::Lena"]}, "source_reports": {"agent_a": {"positive": [], "negative": ["Mia does not has advising clearance."]}, "agent_b": {"positive": ["Lena passed the qualifying exam."], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "explicit_negative", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_advising_clearance.\nAgent B open/incomplete predicates: has_external_scholarship, may_register, needs_manual_audit, passed_exam.\n\nPredicate glossary:\n- has_advising_clearance: has advising clearance\n- has_external_scholarship: has an external scholarship\n- may_register: may register for the seminar\n- needs_manual_audit: needs a manual audit\n- passed_exam: passed the qualifying exam\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- Mia does not has advising clearance.\n\nAgent B report:\nPositive facts:\n- Lena passed the qualifying exam.\nNegative facts:\n- none\n\nTarget atom: has_advising_clearance::Mia.\nTarget statement: Mia has advising clearance.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "university_registration_0072_explicit_negative_closed__owa__multi_agent", "source_id": "university_registration_0072_explicit_negative_closed__owa", "base_id": "university_registration_0072_explicit_negative_closed", "split": "test", "domain": "university_registration", "family": "explicit_negative_closed", "subset": "control_explicit_negative", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["has_advising_clearance", "has_external_scholarship", "may_register", "needs_manual_audit", "passed_exam"], "facts_positive": ["Lena passed the qualifying exam."], "facts_negative": ["Mia does not has advising clearance."], "rules_natural": [], "predicate_glossary": {"passed_exam": "passed the qualifying exam", "has_advising_clearance": "has advising clearance", "has_external_scholarship": "has an external scholarship", "may_register": "may register for the seminar", "needs_manual_audit": "needs a manual audit"}, "vocabulary_predicates": ["has_advising_clearance", "has_external_scholarship", "may_register", "needs_manual_audit", "passed_exam"], "query_atom": "has_advising_clearance::Mia", "target_statement": "Mia has advising clearance.", "gold_reason_type": "explicit_negative_fact", "symbolic": {"positive_atoms": ["passed_exam::Lena"], "negative_atoms": ["has_advising_clearance::Mia"], "rules": [], "query_atom": "has_advising_clearance::Mia", "closure_atoms": ["passed_exam::Lena"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Lena passed the qualifying exam."], "negative": ["Mia does not has advising clearance."]}}, "gold_truth_value": "false", "gold_source_used": "agent_b", "gold_closure_handling": "explicit_negative", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: has_advising_clearance, has_external_scholarship, may_register, needs_manual_audit, passed_exam.\n\nPredicate glossary:\n- has_advising_clearance: has advising clearance\n- has_external_scholarship: has an external scholarship\n- may_register: may register for the seminar\n- needs_manual_audit: needs a manual audit\n- passed_exam: passed the qualifying exam\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Lena passed the qualifying exam.\nNegative facts:\n- Mia does not has advising clearance.\n\nTarget atom: has_advising_clearance::Mia.\nTarget statement: Mia has advising clearance.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "university_registration_0075_explicit_negative_closed__cwa__multi_agent", "source_id": "university_registration_0075_explicit_negative_closed__cwa", "base_id": "university_registration_0075_explicit_negative_closed", "split": "train", "domain": "university_registration", "family": "explicit_negative_closed", "subset": "control_explicit_negative", "semantics": "cwa", "semantic_contract": {"name": "Closed-world assumption", "instruction": "All predicates in the vocabulary are complete. After applying the rules, every unstated and underivable atom is false."}, "closed_predicates": ["has_advising_clearance", "has_external_scholarship", "may_register", "needs_manual_audit", "passed_exam"], "agent_a_complete_predicates": ["has_advising_clearance", "has_external_scholarship", "may_register", "needs_manual_audit", "passed_exam"], "agent_b_open_predicates": [], "facts_positive": ["Theo passed the qualifying exam."], "facts_negative": ["Omar does not has advising clearance."], "rules_natural": [], "predicate_glossary": {"passed_exam": "passed the qualifying exam", "has_advising_clearance": "has advising clearance", "has_external_scholarship": "has an external scholarship", "may_register": "may register for the seminar", "needs_manual_audit": "needs a manual audit"}, "vocabulary_predicates": ["has_advising_clearance", "has_external_scholarship", "may_register", "needs_manual_audit", "passed_exam"], "query_atom": "has_advising_clearance::Omar", "target_statement": "Omar has advising clearance.", "gold_reason_type": "explicit_negative_fact", "symbolic": {"positive_atoms": ["passed_exam::Theo"], "negative_atoms": ["has_advising_clearance::Omar"], "rules": [], "query_atom": "has_advising_clearance::Omar", "closure_atoms": ["passed_exam::Theo"]}, "source_reports": {"agent_a": {"positive": ["Theo passed the qualifying exam."], "negative": ["Omar does not has advising clearance."]}, "agent_b": {"positive": [], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "explicit_negative", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_advising_clearance, has_external_scholarship, may_register, needs_manual_audit, passed_exam.\nAgent B open/incomplete predicates: none.\n\nPredicate glossary:\n- has_advising_clearance: has advising clearance\n- has_external_scholarship: has an external scholarship\n- may_register: may register for the seminar\n- needs_manual_audit: needs a manual audit\n- passed_exam: passed the qualifying exam\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- Theo passed the qualifying exam.\nNegative facts:\n- Omar does not has advising clearance.\n\nAgent B report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nTarget atom: has_advising_clearance::Omar.\nTarget statement: Omar has advising clearance.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "university_registration_0075_explicit_negative_closed__lcwa__multi_agent", "source_id": "university_registration_0075_explicit_negative_closed__lcwa", "base_id": "university_registration_0075_explicit_negative_closed", "split": "train", "domain": "university_registration", "family": "explicit_negative_closed", "subset": "control_explicit_negative", "semantics": "lcwa", "semantic_contract": {"name": "Locally closed-world assumption", "instruction": "Only the predicates listed as complete are closed. After applying the rules, an unstated and underivable atom whose predicate is complete is false; an unstated and underivable atom whose predicate is not complete is unknown."}, "closed_predicates": ["has_advising_clearance"], "agent_a_complete_predicates": ["has_advising_clearance"], "agent_b_open_predicates": ["has_external_scholarship", "may_register", "needs_manual_audit", "passed_exam"], "facts_positive": ["Theo passed the qualifying exam."], "facts_negative": ["Omar does not has advising clearance."], "rules_natural": [], "predicate_glossary": {"passed_exam": "passed the qualifying exam", "has_advising_clearance": "has advising clearance", "has_external_scholarship": "has an external scholarship", "may_register": "may register for the seminar", "needs_manual_audit": "needs a manual audit"}, "vocabulary_predicates": ["has_advising_clearance", "has_external_scholarship", "may_register", "needs_manual_audit", "passed_exam"], "query_atom": "has_advising_clearance::Omar", "target_statement": "Omar has advising clearance.", "gold_reason_type": "explicit_negative_fact", "symbolic": {"positive_atoms": ["passed_exam::Theo"], "negative_atoms": ["has_advising_clearance::Omar"], "rules": [], "query_atom": "has_advising_clearance::Omar", "closure_atoms": ["passed_exam::Theo"]}, "source_reports": {"agent_a": {"positive": [], "negative": ["Omar does not has advising clearance."]}, "agent_b": {"positive": ["Theo passed the qualifying exam."], "negative": []}}, "gold_truth_value": "false", "gold_source_used": "agent_a", "gold_closure_handling": "explicit_negative", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: has_advising_clearance.\nAgent B open/incomplete predicates: has_external_scholarship, may_register, needs_manual_audit, passed_exam.\n\nPredicate glossary:\n- has_advising_clearance: has advising clearance\n- has_external_scholarship: has an external scholarship\n- may_register: may register for the seminar\n- needs_manual_audit: needs a manual audit\n- passed_exam: passed the qualifying exam\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- Omar does not has advising clearance.\n\nAgent B report:\nPositive facts:\n- Theo passed the qualifying exam.\nNegative facts:\n- none\n\nTarget atom: has_advising_clearance::Omar.\nTarget statement: Omar has advising clearance.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}
+{"id": "university_registration_0075_explicit_negative_closed__owa__multi_agent", "source_id": "university_registration_0075_explicit_negative_closed__owa", "base_id": "university_registration_0075_explicit_negative_closed", "split": "train", "domain": "university_registration", "family": "explicit_negative_closed", "subset": "control_explicit_negative", "semantics": "owa", "semantic_contract": {"name": "Open-world assumption", "instruction": "Predicates are open by default. A target atom that is not explicitly stated, explicitly negated, or derivable from the rules is unknown."}, "closed_predicates": [], "agent_a_complete_predicates": [], "agent_b_open_predicates": ["has_advising_clearance", "has_external_scholarship", "may_register", "needs_manual_audit", "passed_exam"], "facts_positive": ["Theo passed the qualifying exam."], "facts_negative": ["Omar does not has advising clearance."], "rules_natural": [], "predicate_glossary": {"passed_exam": "passed the qualifying exam", "has_advising_clearance": "has advising clearance", "has_external_scholarship": "has an external scholarship", "may_register": "may register for the seminar", "needs_manual_audit": "needs a manual audit"}, "vocabulary_predicates": ["has_advising_clearance", "has_external_scholarship", "may_register", "needs_manual_audit", "passed_exam"], "query_atom": "has_advising_clearance::Omar", "target_statement": "Omar has advising clearance.", "gold_reason_type": "explicit_negative_fact", "symbolic": {"positive_atoms": ["passed_exam::Theo"], "negative_atoms": ["has_advising_clearance::Omar"], "rules": [], "query_atom": "has_advising_clearance::Omar", "closure_atoms": ["passed_exam::Theo"]}, "source_reports": {"agent_a": {"positive": [], "negative": []}, "agent_b": {"positive": ["Theo passed the qualifying exam."], "negative": ["Omar does not has advising clearance."]}}, "gold_truth_value": "false", "gold_source_used": "agent_b", "gold_closure_handling": "explicit_negative", "prompt": "You are the coordinator in a multi-agent workflow. Two source agents reported evidence from different systems. Your job is to answer the target statement while preserving source-scoped closure contracts.\nReturn exactly one JSON object with keys truth_value, source_used, closure_handling, and rationale.\n\nAllowed values:\n- truth_value: true, false, unknown.\n- source_used: agent_a, agent_b, both, neither.\n- closure_handling: entailed_or_explicit, explicit_negative, closed_absence, open_absence.\n\nSource attribution rule:\n- Use agent_a when an Agent A complete-source fact, negative fact, or closed absence determines the answer.\n- Use agent_b when an Agent B open-source fact, negative fact, or open absence determines the answer.\n- Use both when a rule-derived answer depends on facts from both source agents.\n- Use neither only if no source report or source-scoped absence is relevant.\n\nSource-scoped closure contract:\n- Agent A is an authoritative system of record only for the listed complete predicates.\n- For Agent A complete predicates, after applying rules, absence of a report means false.\n- Agent B is an incomplete/open source. Absence from Agent B never means false by itself.\n- Do not transfer Agent A completeness to Agent B-only/open predicates.\n- Do not ignore Agent A completeness for predicates it explicitly covers.\n\nAgent A complete predicates: none.\nAgent B open/incomplete predicates: has_advising_clearance, has_external_scholarship, may_register, needs_manual_audit, passed_exam.\n\nPredicate glossary:\n- has_advising_clearance: has advising clearance\n- has_external_scholarship: has an external scholarship\n- may_register: may register for the seminar\n- needs_manual_audit: needs a manual audit\n- passed_exam: passed the qualifying exam\n\nCoordinator policy rules:\n- none\n\nAgent A report:\nPositive facts:\n- none\nNegative facts:\n- none\n\nAgent B report:\nPositive facts:\n- Theo passed the qualifying exam.\nNegative facts:\n- Omar does not has advising clearance.\n\nTarget atom: has_advising_clearance::Omar.\nTarget statement: Omar has advising clearance.\n\nDecide the truth value, identify which source justified the decision, and classify how closure was handled."}