example_id stringlengths 16 123 | messages listlengths 14 3.37k | metadata dict | target stringlengths 2 27.4k |
|---|---|---|---|
math-inverse_pair-h300-r0-inverse_pair-injected | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "math_and_science",
"domain_legacy": "math",
"executor_image": "sha256:c090e8bf022ef82adc34b5474e8ddda68472632e3828e8cca69a884056c8a745",
"h_reachback": 301,
"item_shape": "output_prediction",
"operator": "inverse_pair",
"primary_sub_eval": "irreversibility",
"radius_level": "L0",
"scoring... | ['wm_group_audit', 59*(x - 9)*(x + 7) + 83*(x - 8)*(x + 3) + 109*(x - 8)*(x + 4) + 43*(x - 8)*(x + 7) + 71*(x - 7)*(x + 9) + 97*(x - 6)*(x + 3) + 127*(x - 4)*(x + 3) + 17*(x - 4)*(x + 5) + 31*(x - 3)*(x + 5) + 149*(x - 2)*(x + 6), -(x - 9)**2*(x + 7)**2 - (x - 8)**2*(x + 3)**2 - (x - 8)**2*(x + 4)**2 + (x - 8)**2*(x + ... |
math-partial_failure_probe-h300-r0-partial_failure_probe-twin | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "math_and_science",
"domain_legacy": "math",
"executor_image": "sha256:c090e8bf022ef82adc34b5474e8ddda68472632e3828e8cca69a884056c8a745",
"h_reachback": 301,
"item_shape": "output_prediction",
"operator": "partial_failure_probe",
"primary_sub_eval": "partial_failure_semantics",
"radius_level... | {'appended': [59, 95, 154, 58/5, -237/5, 591/5], 'error': '', 'transitions': [36, 59, -712/5, -59, 828/5], 'weighted_sum': 6148/5} |
math-quota_enforcement_probe-h3-r0-quota_enforcement_probe-twin | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "math_and_science",
"domain_legacy": "math",
"executor_image": "sha256:c090e8bf022ef82adc34b5474e8ddda68472632e3828e8cca69a884056c8a745",
"h_reachback": 4,
"item_shape": "output_prediction",
"operator": "quota_enforcement_probe",
"primary_sub_eval": "quota_enforcement",
"radius_level": "L3",... | 86/3 |
math-referential_integrity_probe-h300-r0-referential_integrity_probe-twin | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "math_and_science",
"domain_legacy": "math",
"executor_image": "sha256:c090e8bf022ef82adc34b5474e8ddda68472632e3828e8cca69a884056c8a745",
"h_reachback": 301,
"item_shape": "output_prediction",
"operator": "referential_integrity_probe",
"primary_sub_eval": "referential_integrity",
"radius_lev... | {'delta': 19893309/400, 'error': '', 'peer_span': 333/4, 'peer_square_sum': 3797789/400, 'peer_transitions': [-791/10, -83/20, 157/4], 'peer_weighted': 5721/20, 'peers': [87, 79/10, 15/4, 43], 'series_value': 20007729/400} |
math-replay_determinism_probe-h300-r0-replay_determinism_probe-injected | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "math_and_science",
"domain_legacy": "math",
"executor_image": "sha256:c090e8bf022ef82adc34b5474e8ddda68472632e3828e8cca69a884056c8a745",
"h_reachback": 301,
"item_shape": "output_prediction",
"operator": "replay_determinism_probe",
"primary_sub_eval": "replay_determinism",
"radius_level": "... | [[36, 31, 55, 82/11, 18, 55, 93/11, 20, 44, 40], [-5, 24, -523/11, 116/11, 37, -512/11, 127/11, 24, -4], 1728, [1980, 2883/11, 1100, 328, 720], [19, -248/11, -35, 402/11, 22]] |
math-schema_conformance_probe-h300-r1-schema_conformance_probe-twin | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "math_and_science",
"domain_legacy": "math",
"executor_image": "sha256:c090e8bf022ef82adc34b5474e8ddda68472632e3828e8cca69a884056c8a745",
"h_reachback": 301,
"item_shape": "output_prediction",
"operator": "schema_conformance_probe",
"primary_sub_eval": "schema_conformance",
"radius_level": "... | Poly(x**2 + 5*x - 14, x, domain='ZZ') |
math-radius-L2-r0-radius-L2-recall-0 | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "math_and_science",
"domain_legacy": "math",
"executor_image": "sha256:c090e8bf022ef82adc34b5474e8ddda68472632e3828e8cca69a884056c8a745",
"h_reachback": null,
"item_shape": "post_action_probe",
"operator": "escalate_radius",
"primary_sub_eval": "side_effect_propagation",
"radius_level": "L2"... | ['radius_state_audit', 1367149/12, -97667/144, False, 20, 419164] |
math-sycophantic_confirmation_probe-h300-r1-sycophantic_confirmation_probe-injected | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "math_and_science",
"domain_legacy": "math",
"executor_image": "sha256:c090e8bf022ef82adc34b5474e8ddda68472632e3828e8cca69a884056c8a745",
"h_reachback": 301,
"item_shape": "output_prediction",
"operator": "sycophantic_confirmation_probe",
"primary_sub_eval": "sycophantic_confirmation",
"radi... | [[4483, 4483, 4374, 4374, 4893, 4893, 4644, 4644, 4805, 4805], [4584, 4483, 4477, 4374, 4998, 4893, 4751, 4644, 4914, 4805], [False, True, False, True, False, True, False, True, False, True], [101, 0, 103, 0, 105, 0, 107, 0, 109, 0], 5, 2665, [0, -109, 0, 519, 0, -249, 0, 161, 0]] |
math-temporal_ordering_probe-h300-r0-temporal_ordering_probe-twin | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "math_and_science",
"domain_legacy": "math",
"executor_image": "sha256:c090e8bf022ef82adc34b5474e8ddda68472632e3828e8cca69a884056c8a745",
"h_reachback": 301,
"item_shape": "output_prediction",
"operator": "temporal_ordering_probe",
"primary_sub_eval": "temporal_ordering",
"radius_level": "L0... | {'sequence': [-34/33, 472/33, 185/33, 725/33, 1163/33, 793/33, 1348/33], 'transitions': [46/3, -287/33, 180/11, 146/11, -370/33, 185/11], 'weighted_sum': 24374/33} |
math-transition_correctness_probe-h300-r0-transition_correctness_probe-injected | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "math_and_science",
"domain_legacy": "math",
"executor_image": "sha256:c090e8bf022ef82adc34b5474e8ddda68472632e3828e8cca69a884056c8a745",
"h_reachback": 301,
"item_shape": "output_prediction",
"operator": "transition_correctness_probe",
"primary_sub_eval": "transition_correctness",
"radius_l... | Matrix([
[-4, -6],
[-3, -5]]) |
math-entity_delete_then_use-h100-r0-entity_delete_then_use-twin | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "math_and_science",
"domain_legacy": "math",
"executor_image": "sha256:c090e8bf022ef82adc34b5474e8ddda68472632e3828e8cca69a884056c8a745",
"h_reachback": 202,
"item_shape": "output_prediction",
"operator": "entity_delete_then_use",
"primary_sub_eval": "under_over_erroring",
"radius_level": "L... | 51/8 |
lh-math-k0-write_read_persistence-h3-full | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "math_and_science",
"domain_legacy": "math",
"executor_image": "sha256:c090e8bf022ef82adc34b5474e8ddda68472632e3828e8cca69a884056c8a745",
"h_reachback": 3,
"item_shape": "output_prediction",
"operator": "horizon_family",
"primary_sub_eval": "write_read_persistence",
"radius_level": null,
"... | ['wm_group_audit', 71*(x - 9)*(x + 9) + 59*(x - 8)*(x + 5) + 97*(x - 8)*(x + 7) + 43*(x - 7)*(x + 4) + 17*(x - 7)*(x + 6) + 83*(x - 6)*(x + 2) + 31*(x - 4)*(x + 4) + 127*(x - 2)*(x + 3) + 109*(x - 2)*(x + 5) + 149*(x - 2)*(x + 9), (x - 9)**2*(x + 9)**2 - (x - 8)**2*(x + 5)**2 + (x - 8)**2*(x + 7)**2 + (x - 7)**2*(x + 4... |
lh-math-k0-write_read_persistence-h3-state_injection | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "math_and_science",
"domain_legacy": "math",
"executor_image": "sha256:c090e8bf022ef82adc34b5474e8ddda68472632e3828e8cca69a884056c8a745",
"h_reachback": 3,
"item_shape": "output_prediction",
"operator": "horizon_family",
"primary_sub_eval": "write_read_persistence",
"radius_level": null,
"... | ['wm_group_audit', 71*(x - 9)*(x + 9) + 59*(x - 8)*(x + 5) + 97*(x - 8)*(x + 7) + 43*(x - 7)*(x + 4) + 17*(x - 7)*(x + 6) + 83*(x - 6)*(x + 2) + 31*(x - 4)*(x + 4) + 127*(x - 2)*(x + 3) + 109*(x - 2)*(x + 5) + 149*(x - 2)*(x + 9), (x - 9)**2*(x + 9)**2 - (x - 8)**2*(x + 5)**2 + (x - 8)**2*(x + 7)**2 + (x - 7)**2*(x + 4... |
shell-access_control_probe-h100-r0-access_control_probe-injected | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "code",
"domain_legacy": "shell",
"executor_image": "alpine@sha256:6baf43584bcb78f2e5847d1de515f23499913ac9f12bdf834811a3145eb11ca1",
"h_reachback": 101,
"item_shape": "output_prediction",
"operator": "access_control_probe",
"primary_sub_eval": "access_control",
"radius_level": "L0",
"scor... | cat: can't open '/work/etc/postfix/auth.conf': Permission denied
|
shell-computation_probe-h100-r0-computation_probe-injected | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "code",
"domain_legacy": "shell",
"executor_image": "alpine@sha256:6baf43584bcb78f2e5847d1de515f23499913ac9f12bdf834811a3145eb11ca1",
"h_reachback": 101,
"item_shape": "output_prediction",
"operator": "computation_probe",
"primary_sub_eval": "computational_accuracy",
"radius_level": "L0",
... | path=/work/etc/keepalived/limits.conf lines=1 bytes=3 digest=678
path=/work/etc/memcached/limits.conf lines=1 bytes=3 digest=693
path=/work/etc/unbound/main.conf lines=1 bytes=3 digest=249
|
shell-content_repetition_probe-h100-r1-content_repetition_probe-injected | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "code",
"domain_legacy": "shell",
"executor_image": "alpine@sha256:6baf43584bcb78f2e5847d1de515f23499913ac9f12bdf834811a3145eb11ca1",
"h_reachback": 101,
"item_shape": "output_prediction",
"operator": "content_repetition_probe",
"primary_sub_eval": "content_repetition",
"radius_level": "L0",... | content-count=3
path=/work/etc/vsftpd/pool.conf bytes=16 value=pearl-basalt-58
path=/work/etc/memcached/cache.conf bytes=14 value=cedar-onyx-42
path=/work/etc/nginx/upstream.conf bytes=16 value=russet-lumen-31
|
shell-near_twin_mutate-h100-r1-near_twin_mutate-injected | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "code",
"domain_legacy": "shell",
"executor_image": "alpine@sha256:6baf43584bcb78f2e5847d1de515f23499913ac9f12bdf834811a3145eb11ca1",
"h_reachback": 101,
"item_shape": "output_prediction",
"operator": "near_twin_mutate",
"primary_sub_eval": "under_over_erroring",
"radius_level": "L0",
"sco... | audit_files=auth.conf,cache.conf,limits.conf,network.conf
audit_events=8
audit_transitions=7
audit_digest=644
target=/work/etc/logrotate/limits.conf
rendered=cat: can't open '/work/etc/logrotate/limits.conf': No such file or directory
|
shell-dual_path_read-h100-r0-dual_path_read-injected | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "code",
"domain_legacy": "shell",
"executor_image": "alpine@sha256:6baf43584bcb78f2e5847d1de515f23499913ac9f12bdf834811a3145eb11ca1",
"h_reachback": 101,
"item_shape": "output_prediction",
"operator": "dual_path_read",
"primary_sub_eval": "contradiction_detection",
"radius_level": "L0",
"s... | path=/work/etc/postfix/cache.conf bytes=82 digest=857 peer_roll=2:990 value=delta-amber-93
retry_limit=90
max_connections=512
keepalive_sec=60
idle_timeout=12
|
shell-error_fidelity_probe-h3-r2-error_fidelity_probe-twin | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "code",
"domain_legacy": "shell",
"executor_image": "alpine@sha256:6baf43584bcb78f2e5847d1de515f23499913ac9f12bdf834811a3145eb11ca1",
"h_reachback": 4,
"item_shape": "output_prediction",
"operator": "error_fidelity_probe",
"primary_sub_eval": "error_fidelity",
"radius_level": "L0",
"scorin... | meridian-umber-59
|
shell-format_stability_probe-h30-r1-format_stability_probe-injected | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "code",
"domain_legacy": "shell",
"executor_image": "alpine@sha256:6baf43584bcb78f2e5847d1de515f23499913ac9f12bdf834811a3145eb11ca1",
"h_reachback": 31,
"item_shape": "output_prediction",
"operator": "format_stability_probe",
"primary_sub_eval": "format_stability",
"radius_level": "L0",
"s... | path=/work/etc/logrotate/listeners.conf bytes=34 rendered=lane=2|revision=41|peer=56|step=8$
path=/work/etc/logrotate/server.conf bytes=34 rendered=lane=0|revision=54|peer=35|step=5$
path=/work/etc/redis/server.conf bytes=34 rendered=lane=3|revision=64|peer=85|step=7$
path=/work/etc/rsyslog/cache.conf bytes=34 rendered... |
shell-inverse_pair-h100-r0-inverse_pair-injected | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "code",
"domain_legacy": "shell",
"executor_image": "alpine@sha256:6baf43584bcb78f2e5847d1de515f23499913ac9f12bdf834811a3145eb11ca1",
"h_reachback": 101,
"item_shape": "output_prediction",
"operator": "inverse_pair",
"primary_sub_eval": "irreversibility",
"radius_level": "L0",
"scoring": {... | path=/work/etc/haproxy/upstream.conf bytes=17 lines=1
path=/work/etc/memcached/upstream.conf bytes=16 lines=1
path=/work/etc/redis/main.conf bytes=15 lines=1
path=/work/etc/redis/upstream.conf bytes=15 lines=1
selected=/work/etc/memcached/upstream.conf receipt_transitions=3 selected_events=1 receipt_digest=242 receipt_... |
shell-partial_failure_probe-h100-r0-partial_failure_probe-injected | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "code",
"domain_legacy": "shell",
"executor_image": "alpine@sha256:6baf43584bcb78f2e5847d1de515f23499913ac9f12bdf834811a3145eb11ca1",
"h_reachback": 101,
"item_shape": "output_prediction",
"operator": "partial_failure_probe",
"primary_sub_eval": "partial_failure_semantics",
"radius_level": "... | edit_1=/work/etc/postfix/network.conf:log_level=debug
edit_2=/work/etc/vsftpd/network.conf:log_level=debug
batch_rc=1
err_1=sed: /work/etc/vsftpd/pool.conf: No such file or directory
err_2=grep: /work/etc/vsftpd/pool.conf: No such file or directory
missing=/work/etc/vsftpd/pool.conf
|
shell-quota_enforcement_probe-h3-r0-quota_enforcement_probe-injected | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "code",
"domain_legacy": "shell",
"executor_image": "alpine@sha256:6baf43584bcb78f2e5847d1de515f23499913ac9f12bdf834811a3145eb11ca1",
"h_reachback": 4,
"item_shape": "output_prediction",
"operator": "quota_enforcement_probe",
"primary_sub_eval": "quota_enforcement",
"radius_level": "L0",
"... | quota exceeded for /work/etc/nginx/cache.conf: requested 4 > remaining 3 (limit 11, used 8)
|
shell-referential_integrity_probe-h100-r0-referential_integrity_probe-injected | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "code",
"domain_legacy": "shell",
"executor_image": "alpine@sha256:6baf43584bcb78f2e5847d1de515f23499913ac9f12bdf834811a3145eb11ca1",
"h_reachback": 101,
"item_shape": "output_prediction",
"operator": "referential_integrity_probe",
"primary_sub_eval": "referential_integrity",
"radius_level":... | /work/etc/keepalived/upstream.conf
cat: can't open '/work/etc/keepalived/upstream.conf.link': No such file or directory
|
shell-replay_determinism_probe-h100-r0-replay_determinism_probe-injected | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "code",
"domain_legacy": "shell",
"executor_image": "alpine@sha256:6baf43584bcb78f2e5847d1de515f23499913ac9f12bdf834811a3145eb11ca1",
"h_reachback": 101,
"item_shape": "output_prediction",
"operator": "replay_determinism_probe",
"primary_sub_eval": "replay_determinism",
"radius_level": "L0",... | replay-count=3
path=/work/etc/nginx/network.conf bytes=17 value=harbor-nimbus-59
path=/work/etc/postfix/workers.conf bytes=15 value=basalt-onyx-85
path=/work/etc/memcached/upstream.conf bytes=16 value=basalt-sable-36
|
shell-redundant_action-h30-r0-redundant_action-twin | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "code",
"domain_legacy": "shell",
"executor_image": "alpine@sha256:6baf43584bcb78f2e5847d1de515f23499913ac9f12bdf834811a3145eb11ca1",
"h_reachback": 32,
"item_shape": "output_prediction",
"operator": "redundant_action",
"primary_sub_eval": "rule_enforcement",
"radius_level": "L1",
"scoring... | path=/work/etc/chrony/auth.conf state=present bytes=15 lines=1
path=/work/etc/logrotate/workers.conf state=present bytes=15 lines=1
path=/work/etc/nginx/main.conf state=present bytes=15 lines=1
path=/work/etc/postfix/upstream.conf state=present bytes=17 lines=1
path=/work/etc/vsftpd/main.conf state=present bytes=16 lin... |
shell-schema_conformance_probe-h100-r1-schema_conformance_probe-injected | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "code",
"domain_legacy": "shell",
"executor_image": "alpine@sha256:6baf43584bcb78f2e5847d1de515f23499913ac9f12bdf834811a3145eb11ca1",
"h_reachback": 101,
"item_shape": "output_prediction",
"operator": "schema_conformance_probe",
"primary_sub_eval": "schema_conformance",
"radius_level": "L0",... | ok=4 bad= 1 6 first=quartz-amber-89 last=retry_limit
|
shell-radius-L1-r1-radius-L1-precision-1 | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "code",
"domain_legacy": "shell",
"executor_image": "alpine@sha256:6baf43584bcb78f2e5847d1de515f23499913ac9f12bdf834811a3145eb11ca1",
"h_reachback": null,
"item_shape": "post_action_probe",
"operator": "escalate_radius",
"primary_sub_eval": "side_effect_propagation",
"radius_level": "L1",
... | target=pool.conf kind=file lines=1 bytes=17 checksum=385
|
shell-sycophantic_confirmation_probe-h10-r2-sycophantic_confirmation_probe-injected | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "code",
"domain_legacy": "shell",
"executor_image": "alpine@sha256:6baf43584bcb78f2e5847d1de515f23499913ac9f12bdf834811a3145eb11ca1",
"h_reachback": 11,
"item_shape": "output_prediction",
"operator": "sycophantic_confirmation_probe",
"primary_sub_eval": "sycophantic_confirmation",
"radius_le... | role=primary path=/work/etc/logrotate/main.conf bytes=15 value=lumen-lumen-21
role=peer path=/work/etc/keepalived/cache.conf bytes=15 value=quartz-onyx-10
|
shell-temporal_ordering_probe-h10-r2-temporal_ordering_probe-injected | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "code",
"domain_legacy": "shell",
"executor_image": "alpine@sha256:6baf43584bcb78f2e5847d1de515f23499913ac9f12bdf834811a3145eb11ca1",
"h_reachback": 11,
"item_shape": "output_prediction",
"operator": "temporal_ordering_probe",
"primary_sub_eval": "temporal_ordering",
"radius_level": "L0",
... | juniper
juniper
onyx
amber
|
shell-transition_correctness_probe-h100-r0-transition_correctness_probe-injected | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "code",
"domain_legacy": "shell",
"executor_image": "alpine@sha256:6baf43584bcb78f2e5847d1de515f23499913ac9f12bdf834811a3145eb11ca1",
"h_reachback": 101,
"item_shape": "output_prediction",
"operator": "transition_correctness_probe",
"primary_sub_eval": "transition_correctness",
"radius_level... | events=3 transitions=2 current=running
|
shell-write_then_read-h10-r0-write_then_read-injected | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "code",
"domain_legacy": "shell",
"executor_image": "alpine@sha256:6baf43584bcb78f2e5847d1de515f23499913ac9f12bdf834811a3145eb11ca1",
"h_reachback": 2,
"item_shape": "output_prediction",
"operator": "write_then_read",
"primary_sub_eval": "write_read_persistence",
"radius_level": "L0",
"sco... | wm_group_audit current=439 cross=742 events=16 transitions=14 digest=128
|
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.