Add files using upload-large-folder tool

.hfignore

@@ -0,0 +1,6 @@
+.cache/
+__pycache__/
+*.pyc
+*.pyo
+round_01_aligned_mix_800/assets/
+persona_aligned_mix_200/assets/

checksums.sha256

@@ -1,13 +1,19 @@
 998b97b25746f2499bf337b9a5fb79b92f9bbcad2f8ef1b0b9efb693e2956f40  .gitattributes
 5b56373d7af94293eff83e888fdbfe1f40ebdb41d2724c3475671e10e3d86070  .gitignore
+5b56373d7af94293eff83e888fdbfe1f40ebdb41d2724c3475671e10e3d86070  .hfignore
 4e7ce122f89dff36d79ce68019940fb84c4824862e07381955b603364a421beb  LICENSE
 e38b31e329d93ff3d4c968be7750d78510c3e390f9c7cfb40af8ca6fc3e74257  README.md
+5491c72eac86410af21932ab6f79c8f5393591db1dc30426b48973f8f978ad3c  croissant/clawbenchpro_collection_croissant.json
+6033bc0fa92fe4b51ac8cce150cd8e2cdf933123b4af7c90847727d264d2c008  croissant/persona_aligned_mix_200_croissant.json
+47d3455506d3f4525c855be7365f843f246ccf94e968889715fa46301ce4d0d6  croissant/round_01_aligned_mix_800_croissant.json
 aa1265325f58610ba9864647d132772414a9c6b594dc5e9b688c2a6d316b0f43  dataset_index.csv
 c40acaaf392927804d85b7496afeb13135205dfbd8828c562b1c6866431ff151  dataset_index.jsonl
-c7382a5db6fffa585f8e1b7f67ed4c45826146dab1ffe1b8ed9b2ebbf0a4e9b5  manifest.json
+9fdb6a48bb584cd1c7ebf16a02d78f8fdb8023abddb8cdcd9ad6f0ed0f6a5a79  manifest.json
 011d2f2a47ada6ad72d66bd8fd275ea7a8ee6ef8301fdede1f6b9eef2400297d  materialize_assets.py
 1d2f56b5d90ce106996e79ff7563c08d4855bcc67924c2ebc67b77b6eb50d14e  persona_aligned_mix_200/README.md
-993fc2a42337a4961a8dfebd620bd817fa46456cd782826fcebb75f6a1afab0d  persona_aligned_mix_200/checksums.sha256
+bc557847f6a9703665bee383921cde25d8ee4a0c771286686935aa5c7fa0b9bb  persona_aligned_mix_200/checksums.sha256
+290ad00b0edd973e53a283bdbcdeb260f8e6bb8183b87f91c70468af897a6f1b  persona_aligned_mix_200/dataset_index.csv
+050657330cc4a20b3261f7668de8cbef1b68b585072fa29cd1dbfd1d21254651  persona_aligned_mix_200/dataset_index.jsonl
 1fb293c7cc9ced181cddd959545abaf99a9a3c56fab8dd826131bd757b5064ad  persona_aligned_mix_200/eval_manifests/base.jsonl
 72aeea0c6321b55982263dbd1cbc23ff114768b3cc21b3cfeab7ff70b7e00284  persona_aligned_mix_200/eval_manifests/base.task_ids
 ddc9ea1b1b06f971daa332d2864fa4a1643dc377693181f75897d2c56056e9af  persona_aligned_mix_200/eval_manifests/hard.jsonl
@@ -1197,7 +1203,9 @@ b675f14fb3a5026397130b64cd4e2ba4f7e76fa360e34ad85d5838b981d5ad32  persona_aligne
 93002945e85af62715331c5abbdd6509430ff1ae5ca8428f0ae1ede1f38624bd  persona_aligned_mix_200/tasks/prompts/data_persona_aligned_skills_50_0049.md
 539d7f3471c7a25cce4874b93d01f7a03cf162fbea98c16182fa8a838107f5f5  persona_aligned_mix_200/tasks/prompts/data_persona_aligned_skills_50_0050.md
 949f64474cc36d32db86c9549f9bcfc37eeff71b5fac7ba7525f7a6c1a9c635c  round_01_aligned_mix_800/README.md
-18eb1b56d5f6e5464e54c795070c5632c9de79df80947b1c88fad85c4171b60d  round_01_aligned_mix_800/checksums.sha256
+c535c736ea6fc905fe828ef5edea92ff549142e58bf9111dc2cc6cd452502a34  round_01_aligned_mix_800/checksums.sha256
+5412d7651f522aa5fc56e96451b386ad993921584b51a8d6b1033995e117461d  round_01_aligned_mix_800/dataset_index.csv
+df9498cec122a957fb01c1b00eb56a2cfefd78137ebe286ec9c317409f21718b  round_01_aligned_mix_800/dataset_index.jsonl
 de93a198234357cef7b5f208ddd351b6e90515c3069ee84b65a12efc8b57e4c4  round_01_aligned_mix_800/eval_manifests/base.jsonl
 82ca2cbf03ffb703078df15a5a2b7832a66a9bd57649d4f9916b86131b06aacf  round_01_aligned_mix_800/eval_manifests/base.task_ids
 9705c9ce482ceb9131b35b8c7fe21230d0df400113e1fe2179ab394c3bc789d8  round_01_aligned_mix_800/eval_manifests/hard_aligned.jsonl

croissant/clawbenchpro_collection_croissant.json

@@ -0,0 +1,255 @@
+{
+  "@context": {
+    "@language": "en",
+    "sc": "https://schema.org/",
+    "cr": "http://mlcommons.org/croissant/",
+    "rai": "http://mlcommons.org/croissant/RAI/",
+    "prov": "http://www.w3.org/ns/prov#"
+  },
+  "@type": "sc:Dataset",
+  "name": "ClawBenchPro",
+  "description": "A compact, builder-based workplace-agent benchmark collection containing round_01_aligned_mix_800 and persona_aligned_mix_200.",
+  "url": "https://huggingface.co/datasets/ErenJaegerYeager/ClawBenchPro",
+  "license": "https://spdx.org/licenses/MIT.html",
+  "conformsTo": "http://mlcommons.org/croissant/1.1",
+  "version": "1.0.0",
+  "datePublished": "2026-05-06",
+  "creator": {
+    "@type": "sc:Organization",
+    "name": "Nanoclaw contributors"
+  },
+  "keywords": [
+    "agent benchmark",
+    "workplace tasks",
+    "tool use",
+    "multi-turn",
+    "skills",
+    "synthetic data"
+  ],
+  "distribution": [
+    {
+      "@type": "cr:FileObject",
+      "@id": "huggingface_repository",
+      "name": "ClawBenchPro Hugging Face repository",
+      "contentUrl": "https://huggingface.co/datasets/ErenJaegerYeager/ClawBenchPro",
+      "encodingFormat": "application/x-directory"
+    },
+    {
+      "@type": "cr:FileObject",
+      "@id": "dataset_index_csv",
+      "name": "dataset_index.csv",
+      "contentUrl": "https://huggingface.co/datasets/ErenJaegerYeager/ClawBenchPro/resolve/main/dataset_index.csv",
+      "encodingFormat": "text/csv"
+    },
+    {
+      "@type": "cr:FileSet",
+      "@id": "round_01_tasks",
+      "name": "round_01_tasks",
+      "containedIn": {
+        "@id": "huggingface_repository"
+      },
+      "includes": "round_01_aligned_mix_800/tasks/**/*",
+      "encodingFormat": "text/*"
+    },
+    {
+      "@type": "cr:FileSet",
+      "@id": "round_01_skills",
+      "name": "round_01_skills",
+      "containedIn": {
+        "@id": "huggingface_repository"
+      },
+      "includes": "round_01_aligned_mix_800/skills/**/*",
+      "encodingFormat": "text/*"
+    },
+    {
+      "@type": "cr:FileSet",
+      "@id": "round_01_eval_manifests",
+      "name": "round_01_eval_manifests",
+      "containedIn": {
+        "@id": "huggingface_repository"
+      },
+      "includes": "round_01_aligned_mix_800/eval_manifests/*",
+      "encodingFormat": "text/*"
+    },
+    {
+      "@type": "cr:FileSet",
+      "@id": "persona_tasks",
+      "name": "persona_tasks",
+      "containedIn": {
+        "@id": "huggingface_repository"
+      },
+      "includes": "persona_aligned_mix_200/tasks/**/*",
+      "encodingFormat": "text/*"
+    },
+    {
+      "@type": "cr:FileSet",
+      "@id": "persona_skills",
+      "name": "persona_skills",
+      "containedIn": {
+        "@id": "huggingface_repository"
+      },
+      "includes": "persona_aligned_mix_200/skills/**/*",
+      "encodingFormat": "text/*"
+    },
+    {
+      "@type": "cr:FileSet",
+      "@id": "persona_eval_manifests",
+      "name": "persona_eval_manifests",
+      "containedIn": {
+        "@id": "huggingface_repository"
+      },
+      "includes": "persona_aligned_mix_200/eval_manifests/*",
+      "encodingFormat": "text/*"
+    }
+  ],
+  "recordSet": [
+    {
+      "@type": "cr:RecordSet",
+      "@id": "task_index",
+      "name": "Task index",
+      "description": "One row per task in ClawBenchPro. This index points to task YAML files, prompt files, task-local builders, generated asset locations, and skill counts.",
+      "field": [
+        {
+          "@type": "cr:Field",
+          "@id": "dataset",
+          "name": "dataset",
+          "dataType": "sc:Text",
+          "source": {
+            "fileObject": {
+              "@id": "dataset_index_csv"
+            },
+            "extract": {
+              "column": "dataset"
+            }
+          }
+        },
+        {
+          "@type": "cr:Field",
+          "@id": "task_id",
+          "name": "task_id",
+          "dataType": "sc:Text",
+          "source": {
+            "fileObject": {
+              "@id": "dataset_index_csv"
+            },
+            "extract": {
+              "column": "task_id"
+            }
+          }
+        },
+        {
+          "@type": "cr:Field",
+          "@id": "category",
+          "name": "category",
+          "dataType": "sc:Text",
+          "source": {
+            "fileObject": {
+              "@id": "dataset_index_csv"
+            },
+            "extract": {
+              "column": "category"
+            }
+          }
+        },
+        {
+          "@type": "cr:Field",
+          "@id": "task_file",
+          "name": "task_file",
+          "dataType": "sc:Text",
+          "source": {
+            "fileObject": {
+              "@id": "dataset_index_csv"
+            },
+            "extract": {
+              "column": "task_file"
+            }
+          }
+        },
+        {
+          "@type": "cr:Field",
+          "@id": "asset_dir",
+          "name": "asset_dir",
+          "dataType": "sc:Text",
+          "source": {
+            "fileObject": {
+              "@id": "dataset_index_csv"
+            },
+            "extract": {
+              "column": "asset_dir"
+            }
+          }
+        },
+        {
+          "@type": "cr:Field",
+          "@id": "prompt_files",
+          "name": "prompt_files",
+          "dataType": "sc:Text",
+          "source": {
+            "fileObject": {
+              "@id": "dataset_index_csv"
+            },
+            "extract": {
+              "column": "prompt_files"
+            }
+          }
+        },
+        {
+          "@type": "cr:Field",
+          "@id": "skill_count",
+          "name": "skill_count",
+          "dataType": "sc:Integer",
+          "source": {
+            "fileObject": {
+              "@id": "dataset_index_csv"
+            },
+            "extract": {
+              "column": "skill_count"
+            }
+          }
+        }
+      ]
+    }
+  ],
+  "rai:dataLimitations": "ClawBenchPro is intended for evaluating workplace-style agent behavior in controlled synthetic tasks. It is not representative of all languages, geographies, industries, accessibility needs, safety-critical domains, or real production workplaces. It should not be used as the sole basis for deployment, hiring, medical, legal, financial, or safety-critical decisions.",
+  "rai:dataBiases": "The tasks were selected and generated to stress agentic workflows, tool use, state tracking, and benchmark coverage. This introduces selection bias toward tasks that are easy to package as local workspaces and may underrepresent low-resource languages, non-technical occupations, and domains requiring embodied or social interaction.",
+  "rai:personalSensitiveInformation": "The dataset is designed as synthetic benchmark data and is not intended to contain real personal data. Some tasks may include fictional personas, synthetic internal URLs, synthetic keys, policy-sensitive text, or domain-specific records as benchmark fixtures.",
+  "rai:dataUseCases": "Validated use cases include comparative evaluation of language-agent runners on workplace-style tasks, category-level benchmark analysis, and reproducibility studies. The dataset is not validated for model fine-tuning, safety certification, demographic fairness auditing, or real-world job-performance measurement.",
+  "rai:dataSocialImpact": "Potential positive impacts include more reproducible agent evaluation and better measurement of multi-turn, tool-using systems. Risks include overgeneralizing benchmark scores, optimizing narrowly to synthetic tasks, or exposing agents to policy-sensitive fixtures without appropriate safeguards. The package documents task categories and uses local synthetic environments to reduce dependence on external services.",
+  "rai:hasSyntheticData": true,
+  "prov:wasDerivedFrom": [
+    {
+      "@id": "nanoclaw_generation_pipeline",
+      "name": "Nanoclaw task generation and import pipeline"
+    }
+  ],
+  "prov:wasGeneratedBy": [
+    {
+      "@type": "prov:Activity",
+      "name": "ClawBenchPro packaging",
+      "description": "Tasks were generated, staged, imported into Nanoclaw task YAML format, validated with env_builder.py builders, grouped into two aligned benchmark datasets, sanitized for local path leakage, and exported in compact builder-only form for hosting.",
+      "prov:used": [
+        "scripts/unpack_task_batch.py",
+        "scripts/import_staged_tasks.py",
+        "scripts/build_published_workplace_datasets.py",
+        "scripts/build_clawbenchpro_hf_dataset.py"
+      ]
+    }
+  ],
+  "additionalProperty": [
+    {
+      "@type": "sc:PropertyValue",
+      "name": "task_count",
+      "value": 1000
+    },
+    {
+      "@type": "sc:PropertyValue",
+      "name": "dataset_root",
+      "value": "."
+    },
+    {
+      "@type": "sc:PropertyValue",
+      "name": "group_counts",
+      "value": "{\"persona_aligned_mix_200\": 200, \"round_01_aligned_mix_800\": 800}"
+    }
+  ]
+}

croissant/persona_aligned_mix_200_croissant.json

@@ -0,0 +1,235 @@
+{
+  "@context": {
+    "@language": "en",
+    "sc": "https://schema.org/",
+    "cr": "http://mlcommons.org/croissant/",
+    "rai": "http://mlcommons.org/croissant/RAI/",
+    "prov": "http://www.w3.org/ns/prov#"
+  },
+  "@type": "sc:Dataset",
+  "name": "persona_aligned_mix_200",
+  "description": "Persona Aligned Mix 200 from the ClawBenchPro benchmark collection.",
+  "url": "https://huggingface.co/datasets/ErenJaegerYeager/ClawBenchPro/tree/main/persona_aligned_mix_200",
+  "license": "https://spdx.org/licenses/MIT.html",
+  "conformsTo": "http://mlcommons.org/croissant/1.1",
+  "version": "1.0.0",
+  "datePublished": "2026-05-06",
+  "creator": {
+    "@type": "sc:Organization",
+    "name": "Nanoclaw contributors"
+  },
+  "keywords": [
+    "agent benchmark",
+    "workplace tasks",
+    "tool use",
+    "multi-turn",
+    "skills",
+    "synthetic data"
+  ],
+  "distribution": [
+    {
+      "@type": "cr:FileObject",
+      "@id": "huggingface_repository",
+      "name": "ClawBenchPro Hugging Face repository",
+      "contentUrl": "https://huggingface.co/datasets/ErenJaegerYeager/ClawBenchPro/tree/main/persona_aligned_mix_200",
+      "encodingFormat": "application/x-directory"
+    },
+    {
+      "@type": "cr:FileObject",
+      "@id": "dataset_index_csv",
+      "name": "dataset_index.csv",
+      "contentUrl": "https://huggingface.co/datasets/ErenJaegerYeager/ClawBenchPro/resolve/main/persona_aligned_mix_200/dataset_index.csv",
+      "encodingFormat": "text/csv"
+    },
+    {
+      "@type": "cr:FileSet",
+      "@id": "persona_aligned_mix_200_tasks",
+      "name": "persona_aligned_mix_200_tasks",
+      "containedIn": {
+        "@id": "huggingface_repository"
+      },
+      "includes": "persona_aligned_mix_200/tasks/**/*",
+      "encodingFormat": "text/*"
+    },
+    {
+      "@type": "cr:FileSet",
+      "@id": "persona_aligned_mix_200_skills",
+      "name": "persona_aligned_mix_200_skills",
+      "containedIn": {
+        "@id": "huggingface_repository"
+      },
+      "includes": "persona_aligned_mix_200/skills/**/*",
+      "encodingFormat": "text/*"
+    },
+    {
+      "@type": "cr:FileSet",
+      "@id": "persona_aligned_mix_200_eval_manifests",
+      "name": "persona_aligned_mix_200_eval_manifests",
+      "containedIn": {
+        "@id": "huggingface_repository"
+      },
+      "includes": "persona_aligned_mix_200/eval_manifests/*",
+      "encodingFormat": "text/*"
+    },
+    {
+      "@type": "cr:FileSet",
+      "@id": "persona_aligned_mix_200_provenance",
+      "name": "persona_aligned_mix_200_provenance",
+      "containedIn": {
+        "@id": "huggingface_repository"
+      },
+      "includes": "persona_aligned_mix_200/provenance/*",
+      "encodingFormat": "text/*"
+    }
+  ],
+  "recordSet": [
+    {
+      "@type": "cr:RecordSet",
+      "@id": "task_index",
+      "name": "Task index",
+      "description": "One row per task in persona_aligned_mix_200. This index points to task YAML files, prompt files, task-local builders, generated asset locations, and skill counts.",
+      "field": [
+        {
+          "@type": "cr:Field",
+          "@id": "dataset",
+          "name": "dataset",
+          "dataType": "sc:Text",
+          "source": {
+            "fileObject": {
+              "@id": "dataset_index_csv"
+            },
+            "extract": {
+              "column": "dataset"
+            }
+          }
+        },
+        {
+          "@type": "cr:Field",
+          "@id": "task_id",
+          "name": "task_id",
+          "dataType": "sc:Text",
+          "source": {
+            "fileObject": {
+              "@id": "dataset_index_csv"
+            },
+            "extract": {
+              "column": "task_id"
+            }
+          }
+        },
+        {
+          "@type": "cr:Field",
+          "@id": "category",
+          "name": "category",
+          "dataType": "sc:Text",
+          "source": {
+            "fileObject": {
+              "@id": "dataset_index_csv"
+            },
+            "extract": {
+              "column": "category"
+            }
+          }
+        },
+        {
+          "@type": "cr:Field",
+          "@id": "task_file",
+          "name": "task_file",
+          "dataType": "sc:Text",
+          "source": {
+            "fileObject": {
+              "@id": "dataset_index_csv"
+            },
+            "extract": {
+              "column": "task_file"
+            }
+          }
+        },
+        {
+          "@type": "cr:Field",
+          "@id": "asset_dir",
+          "name": "asset_dir",
+          "dataType": "sc:Text",
+          "source": {
+            "fileObject": {
+              "@id": "dataset_index_csv"
+            },
+            "extract": {
+              "column": "asset_dir"
+            }
+          }
+        },
+        {
+          "@type": "cr:Field",
+          "@id": "prompt_files",
+          "name": "prompt_files",
+          "dataType": "sc:Text",
+          "source": {
+            "fileObject": {
+              "@id": "dataset_index_csv"
+            },
+            "extract": {
+              "column": "prompt_files"
+            }
+          }
+        },
+        {
+          "@type": "cr:Field",
+          "@id": "skill_count",
+          "name": "skill_count",
+          "dataType": "sc:Integer",
+          "source": {
+            "fileObject": {
+              "@id": "dataset_index_csv"
+            },
+            "extract": {
+              "column": "skill_count"
+            }
+          }
+        }
+      ]
+    }
+  ],
+  "rai:dataLimitations": "ClawBenchPro is intended for evaluating workplace-style agent behavior in controlled synthetic tasks. It is not representative of all languages, geographies, industries, accessibility needs, safety-critical domains, or real production workplaces. It should not be used as the sole basis for deployment, hiring, medical, legal, financial, or safety-critical decisions.",
+  "rai:dataBiases": "The tasks were selected and generated to stress agentic workflows, tool use, state tracking, and benchmark coverage. This introduces selection bias toward tasks that are easy to package as local workspaces and may underrepresent low-resource languages, non-technical occupations, and domains requiring embodied or social interaction.",
+  "rai:personalSensitiveInformation": "The dataset is designed as synthetic benchmark data and is not intended to contain real personal data. Some tasks may include fictional personas, synthetic internal URLs, synthetic keys, policy-sensitive text, or domain-specific records as benchmark fixtures.",
+  "rai:dataUseCases": "Validated use cases include comparative evaluation of language-agent runners on workplace-style tasks, category-level benchmark analysis, and reproducibility studies. The dataset is not validated for model fine-tuning, safety certification, demographic fairness auditing, or real-world job-performance measurement.",
+  "rai:dataSocialImpact": "Potential positive impacts include more reproducible agent evaluation and better measurement of multi-turn, tool-using systems. Risks include overgeneralizing benchmark scores, optimizing narrowly to synthetic tasks, or exposing agents to policy-sensitive fixtures without appropriate safeguards. The package documents task categories and uses local synthetic environments to reduce dependence on external services.",
+  "rai:hasSyntheticData": true,
+  "prov:wasDerivedFrom": [
+    {
+      "@id": "nanoclaw_generation_pipeline",
+      "name": "Nanoclaw task generation and import pipeline"
+    }
+  ],
+  "prov:wasGeneratedBy": [
+    {
+      "@type": "prov:Activity",
+      "name": "ClawBenchPro packaging",
+      "description": "Tasks were generated, staged, imported into Nanoclaw task YAML format, validated with env_builder.py builders, grouped into two aligned benchmark datasets, sanitized for local path leakage, and exported in compact builder-only form for hosting.",
+      "prov:used": [
+        "scripts/unpack_task_batch.py",
+        "scripts/import_staged_tasks.py",
+        "scripts/build_published_workplace_datasets.py",
+        "scripts/build_clawbenchpro_hf_dataset.py"
+      ]
+    }
+  ],
+  "additionalProperty": [
+    {
+      "@type": "sc:PropertyValue",
+      "name": "task_count",
+      "value": 200
+    },
+    {
+      "@type": "sc:PropertyValue",
+      "name": "dataset_root",
+      "value": "persona_aligned_mix_200"
+    },
+    {
+      "@type": "sc:PropertyValue",
+      "name": "group_counts",
+      "value": "{\"base\": 50, \"hard\": 50, \"multi_turn\": 50, \"skills\": 50}"
+    }
+  ]
+}

croissant/round_01_aligned_mix_800_croissant.json

@@ -0,0 +1,235 @@
+{
+  "@context": {
+    "@language": "en",
+    "sc": "https://schema.org/",
+    "cr": "http://mlcommons.org/croissant/",
+    "rai": "http://mlcommons.org/croissant/RAI/",
+    "prov": "http://www.w3.org/ns/prov#"
+  },
+  "@type": "sc:Dataset",
+  "name": "round_01_aligned_mix_800",
+  "description": "Round 01 Aligned Mix 800 from the ClawBenchPro benchmark collection.",
+  "url": "https://huggingface.co/datasets/ErenJaegerYeager/ClawBenchPro/tree/main/round_01_aligned_mix_800",
+  "license": "https://spdx.org/licenses/MIT.html",
+  "conformsTo": "http://mlcommons.org/croissant/1.1",
+  "version": "1.0.0",
+  "datePublished": "2026-05-06",
+  "creator": {
+    "@type": "sc:Organization",
+    "name": "Nanoclaw contributors"
+  },
+  "keywords": [
+    "agent benchmark",
+    "workplace tasks",
+    "tool use",
+    "multi-turn",
+    "skills",
+    "synthetic data"
+  ],
+  "distribution": [
+    {
+      "@type": "cr:FileObject",
+      "@id": "huggingface_repository",
+      "name": "ClawBenchPro Hugging Face repository",
+      "contentUrl": "https://huggingface.co/datasets/ErenJaegerYeager/ClawBenchPro/tree/main/round_01_aligned_mix_800",
+      "encodingFormat": "application/x-directory"
+    },
+    {
+      "@type": "cr:FileObject",
+      "@id": "dataset_index_csv",
+      "name": "dataset_index.csv",
+      "contentUrl": "https://huggingface.co/datasets/ErenJaegerYeager/ClawBenchPro/resolve/main/round_01_aligned_mix_800/dataset_index.csv",
+      "encodingFormat": "text/csv"
+    },
+    {
+      "@type": "cr:FileSet",
+      "@id": "round_01_aligned_mix_800_tasks",
+      "name": "round_01_aligned_mix_800_tasks",
+      "containedIn": {
+        "@id": "huggingface_repository"
+      },
+      "includes": "round_01_aligned_mix_800/tasks/**/*",
+      "encodingFormat": "text/*"
+    },
+    {
+      "@type": "cr:FileSet",
+      "@id": "round_01_aligned_mix_800_skills",
+      "name": "round_01_aligned_mix_800_skills",
+      "containedIn": {
+        "@id": "huggingface_repository"
+      },
+      "includes": "round_01_aligned_mix_800/skills/**/*",
+      "encodingFormat": "text/*"
+    },
+    {
+      "@type": "cr:FileSet",
+      "@id": "round_01_aligned_mix_800_eval_manifests",
+      "name": "round_01_aligned_mix_800_eval_manifests",
+      "containedIn": {
+        "@id": "huggingface_repository"
+      },
+      "includes": "round_01_aligned_mix_800/eval_manifests/*",
+      "encodingFormat": "text/*"
+    },
+    {
+      "@type": "cr:FileSet",
+      "@id": "round_01_aligned_mix_800_provenance",
+      "name": "round_01_aligned_mix_800_provenance",
+      "containedIn": {
+        "@id": "huggingface_repository"
+      },
+      "includes": "round_01_aligned_mix_800/provenance/*",
+      "encodingFormat": "text/*"
+    }
+  ],
+  "recordSet": [
+    {
+      "@type": "cr:RecordSet",
+      "@id": "task_index",
+      "name": "Task index",
+      "description": "One row per task in round_01_aligned_mix_800. This index points to task YAML files, prompt files, task-local builders, generated asset locations, and skill counts.",
+      "field": [
+        {
+          "@type": "cr:Field",
+          "@id": "dataset",
+          "name": "dataset",
+          "dataType": "sc:Text",
+          "source": {
+            "fileObject": {
+              "@id": "dataset_index_csv"
+            },
+            "extract": {
+              "column": "dataset"
+            }
+          }
+        },
+        {
+          "@type": "cr:Field",
+          "@id": "task_id",
+          "name": "task_id",
+          "dataType": "sc:Text",
+          "source": {
+            "fileObject": {
+              "@id": "dataset_index_csv"
+            },
+            "extract": {
+              "column": "task_id"
+            }
+          }
+        },
+        {
+          "@type": "cr:Field",
+          "@id": "category",
+          "name": "category",
+          "dataType": "sc:Text",
+          "source": {
+            "fileObject": {
+              "@id": "dataset_index_csv"
+            },
+            "extract": {
+              "column": "category"
+            }
+          }
+        },
+        {
+          "@type": "cr:Field",
+          "@id": "task_file",
+          "name": "task_file",
+          "dataType": "sc:Text",
+          "source": {
+            "fileObject": {
+              "@id": "dataset_index_csv"
+            },
+            "extract": {
+              "column": "task_file"
+            }
+          }
+        },
+        {
+          "@type": "cr:Field",
+          "@id": "asset_dir",
+          "name": "asset_dir",
+          "dataType": "sc:Text",
+          "source": {
+            "fileObject": {
+              "@id": "dataset_index_csv"
+            },
+            "extract": {
+              "column": "asset_dir"
+            }
+          }
+        },
+        {
+          "@type": "cr:Field",
+          "@id": "prompt_files",
+          "name": "prompt_files",
+          "dataType": "sc:Text",
+          "source": {
+            "fileObject": {
+              "@id": "dataset_index_csv"
+            },
+            "extract": {
+              "column": "prompt_files"
+            }
+          }
+        },
+        {
+          "@type": "cr:Field",
+          "@id": "skill_count",
+          "name": "skill_count",
+          "dataType": "sc:Integer",
+          "source": {
+            "fileObject": {
+              "@id": "dataset_index_csv"
+            },
+            "extract": {
+              "column": "skill_count"
+            }
+          }
+        }
+      ]
+    }
+  ],
+  "rai:dataLimitations": "ClawBenchPro is intended for evaluating workplace-style agent behavior in controlled synthetic tasks. It is not representative of all languages, geographies, industries, accessibility needs, safety-critical domains, or real production workplaces. It should not be used as the sole basis for deployment, hiring, medical, legal, financial, or safety-critical decisions.",
+  "rai:dataBiases": "The tasks were selected and generated to stress agentic workflows, tool use, state tracking, and benchmark coverage. This introduces selection bias toward tasks that are easy to package as local workspaces and may underrepresent low-resource languages, non-technical occupations, and domains requiring embodied or social interaction.",
+  "rai:personalSensitiveInformation": "The dataset is designed as synthetic benchmark data and is not intended to contain real personal data. Some tasks may include fictional personas, synthetic internal URLs, synthetic keys, policy-sensitive text, or domain-specific records as benchmark fixtures.",
+  "rai:dataUseCases": "Validated use cases include comparative evaluation of language-agent runners on workplace-style tasks, category-level benchmark analysis, and reproducibility studies. The dataset is not validated for model fine-tuning, safety certification, demographic fairness auditing, or real-world job-performance measurement.",
+  "rai:dataSocialImpact": "Potential positive impacts include more reproducible agent evaluation and better measurement of multi-turn, tool-using systems. Risks include overgeneralizing benchmark scores, optimizing narrowly to synthetic tasks, or exposing agents to policy-sensitive fixtures without appropriate safeguards. The package documents task categories and uses local synthetic environments to reduce dependence on external services.",
+  "rai:hasSyntheticData": true,
+  "prov:wasDerivedFrom": [
+    {
+      "@id": "nanoclaw_generation_pipeline",
+      "name": "Nanoclaw task generation and import pipeline"
+    }
+  ],
+  "prov:wasGeneratedBy": [
+    {
+      "@type": "prov:Activity",
+      "name": "ClawBenchPro packaging",
+      "description": "Tasks were generated, staged, imported into Nanoclaw task YAML format, validated with env_builder.py builders, grouped into two aligned benchmark datasets, sanitized for local path leakage, and exported in compact builder-only form for hosting.",
+      "prov:used": [
+        "scripts/unpack_task_batch.py",
+        "scripts/import_staged_tasks.py",
+        "scripts/build_published_workplace_datasets.py",
+        "scripts/build_clawbenchpro_hf_dataset.py"
+      ]
+    }
+  ],
+  "additionalProperty": [
+    {
+      "@type": "sc:PropertyValue",
+      "name": "task_count",
+      "value": 800
+    },
+    {
+      "@type": "sc:PropertyValue",
+      "name": "dataset_root",
+      "value": "round_01_aligned_mix_800"
+    },
+    {
+      "@type": "sc:PropertyValue",
+      "name": "group_counts",
+      "value": "{\"base\": 200, \"hard_aligned\": 200, \"multi_turn_aligned\": 200, \"skills_aligned\": 200}"
+    }
+  ]
+}

manifest.json

@@ -1,6 +1,6 @@
 {
   "package_name": "ClawBenchPro",
-  "created_at": "2026-05-06T07:47:40.032231+00:00",
+  "created_at": "2026-05-06T08:17:03.167280+00:00",
   "source_package": "published_datasets/nanoclaw_workplace_suite",
   "format": "huggingface_dataset_repository",
   "license": "MIT",
@@ -16,8 +16,8 @@
         "multi_turn_aligned": 200,
         "skills_aligned": 200
       },
-      "files": 5552,
-      "bytes": 24965910,
+      "files": 5554,
+      "bytes": 25587458,
       "checksums": "round_01_aligned_mix_800/checksums.sha256"
     },
     {
@@ -31,8 +31,8 @@
         "multi_turn": 50,
         "skills": 50
       },
-      "files": 1190,
-      "bytes": 6851989,
+      "files": 1192,
+      "bytes": 7007060,
       "checksums": "persona_aligned_mix_200/checksums.sha256"
     }
   ]

persona_aligned_mix_200/checksums.sha256

@@ -1,4 +1,6 @@
 1d2f56b5d90ce106996e79ff7563c08d4855bcc67924c2ebc67b77b6eb50d14e  README.md
+290ad00b0edd973e53a283bdbcdeb260f8e6bb8183b87f91c70468af897a6f1b  dataset_index.csv
+050657330cc4a20b3261f7668de8cbef1b68b585072fa29cd1dbfd1d21254651  dataset_index.jsonl
 1fb293c7cc9ced181cddd959545abaf99a9a3c56fab8dd826131bd757b5064ad  eval_manifests/base.jsonl
 72aeea0c6321b55982263dbd1cbc23ff114768b3cc21b3cfeab7ff70b7e00284  eval_manifests/base.task_ids
 ddc9ea1b1b06f971daa332d2864fa4a1643dc377693181f75897d2c56056e9af  eval_manifests/hard.jsonl

round_01_aligned_mix_800/checksums.sha256

@@ -1,4 +1,6 @@
 949f64474cc36d32db86c9549f9bcfc37eeff71b5fac7ba7525f7a6c1a9c635c  README.md
+5412d7651f522aa5fc56e96451b386ad993921584b51a8d6b1033995e117461d  dataset_index.csv
+df9498cec122a957fb01c1b00eb56a2cfefd78137ebe286ec9c317409f21718b  dataset_index.jsonl
 de93a198234357cef7b5f208ddd351b6e90515c3069ee84b65a12efc8b57e4c4  eval_manifests/base.jsonl
 82ca2cbf03ffb703078df15a5a2b7832a66a9bd57649d4f9916b86131b06aacf  eval_manifests/base.task_ids
 9705c9ce482ceb9131b35b8c7fe21230d0df400113e1fe2179ab394c3bc789d8  eval_manifests/hard_aligned.jsonl