{
  "schema": 1,
  "created_utc": "2026-11-09T03:45:25.019883+00:00",
  "split": "DEV",
  "dataset_sha256": "db285c0d2086b04c465ca12de91ba29b3025083eb75404e2a476669c2b3f7eb9",
  "evaluator_sha256": "e64363f9fb5fd30417006e9f77ad793bbb07db70d88544309761f56f879cc7c6",
  "runtime_files": {
    "app/__init__.py": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
    "app/main.py": "38ae3f77e8a0ddec414459da1ec17d5247d2f9457bc7ceb2af17d442658a1d28",
    "app/routes.py": "696dfb44e1e1950ac54e73cfa8a154c1567c651e0e5f652b23870f6dfcd01cfe",
    "app/security.py": "ac1dc6c637e007cffb1997007e9b5ec2a398d4719349a9177fe551ca6422e6e0",
    "tools/answers/__init__.py": "8b8ab7f996121e5e5b1b594cf08f4c3b4f16f1b434b74276c803a3a83681d2cb",
    "tools/answers/answer_coverage.py": "746ab3b89ba65f263b63037be55ee02ee25aaaf685dae6e95f7bd6a5536fc6fa",
    "tools/answers/answer_delivery.py": "2520ab327670e35511fb578ef1a447f0c011fb04fa3a341a36696631a2ca499e",
    "tools/answers/answer_errors.py": "d1541b5c31b5d64e0bfa729f5e9a29adddc4b55bdd24508490f39cfc0a44b240",
    "tools/answers/answer_format.py": "533cfe3a30c835d22441d877df63fe08e366d01ed6d15474fe6fcff26bde259b",
    "tools/answers/checked_generation.py": "c96b407f3ed872e9195ebc02a8534df453ebf200fb471d572859b95b36faaaeb",
    "tools/answers/content_selection.py": "b7f7877469f09d3538232e23a4e23add72b40b71ca0f5b9724028f1aa6174fa0",
    "tools/answers/context_policy.py": "526f5de0b9112e23abd76e403799f156d0150c83f03577251af23d1685991327",
    "tools/answers/evidence_selection.py": "1007a85b8838e1eb3c34ee610a4f81dfff57410a90213e02b8ccde90a0fedbd2",
    "tools/answers/graph_answer_prompt.txt": "9f61bf69c8787f9fcf650c0228a30dfed5f18a870c9980c6ef892297ebe6013a",
    "tools/answers/graph_context.py": "5b14150b7f6903cd92c004d56943d949167be8859acf4f97aed14a713e59faff",
    "tools/answers/model_client.py": "f595eb1b594364faff21ce2a6842c8c7e40a5ca4e6a2c3085cdc4a4bae28ae04",
    "tools/answers/presentation.py": "954c65288ddc3c8f873fec3ef99176adb70785806043ac1814c7a01a267b4aad",
    "tools/answers/question_intent.py": "129af1f01371752e7d582cea5b4f21bfa3d7d3742a940854e2e32a93ed63caee",
    "tools/answers/question_plan.py": "9c03d60aed1b5475236c51b8975a83f2d34cc39ce15391c350364a22f9ad4200",
    "tools/answers/rag_evidence.py": "fef7b15fadec291055524f21d2e4db1acade1bae149c629bf15865b2f78d4ed5",
    "tools/answers/rag_preparation.py": "dd7542b9386b338a8c52e7536128ca10edced681f2b653e358e882c5aa98556d",
    "tools/answers/reference_links.py": "eb59206bcb3b79912daafcd591e7adb504ec592e61ea32e4e133cb6a65fe0b57",
    "tools/answers/render_answer.py": "d3fc875436d070f36043952d169d432bbd45bc4d1165112871ce5eaa90000d60",
    "tools/answers/synthesis.py": "a58b08e772bf880ae0a0c540c1da6be281efa935d519b432326ef0e024c42c9d",
    "tools/answers/verified_rag.py": "906df54e11797e1d306605333b114d2b0cc772c49fc15bcd1e23eef92a8dacf1",
    "tools/graphrag/__init__.py": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
    "tools/graphrag/build_graphrag.py": "67c5b994ec1be07eadbeaebfad73aa21fbb88e2a28cee805af0034d7cd160b8b",
    "tools/graphrag/graph_db.py": "5e6364d72cd645087c183633219b4771210c6c3f222946ff4c8e6c512b9b62e5",
    "tools/graphrag/graph_search.py": "96026c31b8df45d84dbf4c7a0066b239a6b3328a9bd9ef312ca9d71dc82aaead",
    "tools/graphrag/graphrag.py": "2ad7fc728761558d294399d52d5e2e66f6c57e1a88dc1aedf4e2e7ff6375fdcc",
    "tools/graphrag/neo4j_smoke.py": "fb5244c72c29ef2acf96d1e7f54335ff168eed4119db1692025eb1b99194b7bb",
    "tools/graphrag/publication.py": "c804b447d02796fb4d53d4b37b843f1ad0d7c259d286b2eb14989ca659184682",
    "tools/knowledge/__init__.py": "905a19d7cf5d95379800b2ad917ad759fa376f23d2f8a244527b9dc187555019",
    "tools/knowledge/academic.py": "af0562ba079b1fa6e2db62bb81c42568567dd749b862a3272407afbea0138865",
    "tools/knowledge/build.py": "4666b1cd2d42533b7289f5e2ecaac11c5cc16e09f3ea4b9a2714d17cb8c2f9f6",
    "tools/knowledge/concepts.py": "de0d005ab298c3d86020b0cb09fc695852303d8a5f411d0a3828b58131713b8d",
    "tools/knowledge/evidence.py": "b0eb7d3e0f6a97083829a9940b3a64dade96d4f98435eb5ebfb43adaf10fac18",
    "tools/knowledge/extraction.py": "c1819afc5072a5999c985b4547809db997ecf3e1112b031534ae34ed295bd106",
    "tools/knowledge/graph_answers.py": "fd599b7ff65485ae11e77022cc007b99e5bd3b533383cc64dc5b7aa3dabf81fa",
    "tools/knowledge/graph_projection.py": "dac3058008fb43c833d239b6a7ced27ef5419004cc768b246741eb2e1a69b12a",
    "tools/knowledge/local_graph.py": "d82fc6d1a42e70d97c1010f4de9a28258727f2fc2cd0e7a9bc6f2c8e51c06c0d",
    "tools/knowledge/planned_answers.py": "5a39356cd9a02d02cbc60bcf670b0cc0648ccf1f16b46b539262dc646080c23f",
    "tools/knowledge/publication.py": "a5a29af22fd73a57f53da47c687200f0ad392d48ce94a23250f00c7a53731148",
    "tools/knowledge/query.py": "a209423b2d068d96a99370208a95e7a15a1fad5c370436a6c9334d6a6200e891",
    "tools/knowledge/sources.py": "709c2201288a920e99578fe4e58e191a742deb5f15170104820680d156280c69",
    "tools/knowledge/store.py": "fa22cd357df8c3cde4cd1b6a4f0a9b78d12721f3970ec896584205f8a5d5d7c1",
    "tools/rag/__init__.py": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
    "tools/rag/build_chunks.py": "037f80a423eba26258a8b1f24e8d9129ce4f043568b7db2574a80d01c5ca32ee",
    "tools/rag/build_sqlite_index.py": "2cb5944543c77d6943fba9f1f7ce8d3c50340bf8e6a4063105e11dbdb608ace0",
    "tools/rag/common.py": "c325b75d50b0cac6f4b780ad0b2a641b8478b9ba6a71680837fabedb947530fe",
    "tools/rag/embedding_cache.py": "4c8faa262c59f85842747ba189fa35c1e4d0a4ce2fa5337bd42f498130022387",
    "tools/rag/embedding_utils.py": "768798b43ca7c99f672ec9e957fb5c3d1c9208409c728cb9af6ae74d31d7d48c",
    "tools/rag/llm_context.py": "9db74da31731948f6ccea9f4b91c2eba206bb22f8757528c21e8b6286c00a340",
    "tools/rag/rag_document_expansion.py": "13ef2db80b6bae517cd46feba6a8096a64399019c5b2dd2a2d5ca1ebcdfbf638",
    "tools/rag/rag_query_evidence.py": "af2dbb7ce94ae625ffc8e01040541b51449b8bd923a4da72027445b779647571",
    "tools/rag/rag_supplemental_expansion.py": "7c26faf29fcc08ba480d7f374d17aa64b355ae87d9cc8bbaee02afb45c54320e",
    "tools/rag/rag_utils.py": "fe42c7f7dc2b1679e0d619b29bd3f087c8e1349da1f45b39b2881eb7ea7f1b4a",
    "tools/rag/rebuild_index.py": "5ee0159f7242d513377824cba6bff8c7d543d4a9c0001dc51499386412f4a3bf",
    "tools/rag/scan_zpwiki.py": "1a24dcd7cc4063b9c398c81619e87775af07a2e0fa5ad824110c528247123e86",
    "tools/rag/search_core.py": "f4757c197c1a1def3f1ab351cef7022968331c8c2ac57345015bcc30c6bfece1",
    "tools/rag/search_db.py": "0356e00c6122b746edee86f28ce86543cfd26e40f55cae31f73feded9be14544",
    "tools/rag/search_scope.py": "360a1f17a3fe341f829dd7685b253dc1ce2e286362431b638d39b1fe468d3f79",
    "tools/rag/search_utils.py": "82ce2567d5837e4d1fa7b40ab91a2f5f7f3c992e33282ef0cf14bc3259110644",
    "tools/rag/sqlite_schema.py": "e2281d500d05891c5c2f6651ca6c116530c8d0bd063637850243f52fc14e2745",
    "tools/retrieval/__init__.py": "32038cbcb49f11f78f5fd439df555322a03996a5d9935c4f2f5a959528161771",
    "tools/retrieval/person_names.py": "6100e3842f5d01f5128929e1cab8b916b8d81295dbff2765b4c73dcb9bb926c2",
    "tools/retrieval/publication.py": "bc5924dcba0dccf7434b277af773c60b212cb521b5a617055d00a456091e4e75",
    "tools/retrieval/question_constraints.py": "d4847b75ab133a8303c5b1b454d9b70bd358dd58c08808c7943d1f31bed44d69",
    "tools/retrieval/retrieval_contract.py": "3478e02c827d169fbecedd6f3cc11218e2636abed5a55bde416e593fc47d5368",
    "tools/retrieval/text_budget.py": "cb676f60e827234ad168df86135871f84a4420c77bc1ccfb8d147879c4b84dcd",
    "tools/retrieval/work_selector.py": "bd89164cbe3a5f9cb04a4c683d1781cfd1ee319c62bf0936421a96748bdaf9c1",
    "tools/runtime/__init__.py": "684ad0cd460df5746f5785326ba7fcdbb017305b2317a96fbd1272d95f6dc2b2",
    "tools/runtime/answer_health.py": "a02f5070d01c691db19337e5278e600fa686ea8bd83fa607620a35b81ece3d1c",
    "tools/runtime/bootstrap_indexes.py": "f032e47390b14138d2f34ae65af68096364fe4eccd1dd3ddc0568806dea3af36",
    "tools/runtime/report_files.py": "f54913b5b2e4465c11c40212b3fe0203cbad84007ffe46281753f485bc9d686d",
    "tools/runtime/runtime_assets.py": "93d9ba92200e96d17f90e7171fff0b43ffe632c05dcfe446838fa87ad84db0bf",
    "tools/runtime/runtime_health.py": "4685a46a0d8468db60fdb7bb3df3dbc1700c724f3dc7226ae431b2bd4b5272e4",
    "tools/runtime/tool_health.py": "bc661bd32a6e6e6694612aa06879d5582a891053175ca6ac7ce2878bcb74701c",
    "tools/sync/__init__.py": "2e0323ee92a26a8487ed7b60abd6c47f8c66755bc695c5e76b845503f34fb1f1",
    "tools/sync/index_state.py": "a29a1daef5e1a0b52a9cfbadd1469d3eb8ab53e25939342b2f2df8941ffaf689",
    "tools/sync/sync_indexes.py": "168d973d514a1220347f28d3fce079ee5a188ac283319cd8cda146a409cae9b8",
    "tools/sync/sync_jobs.py": "d1ec1f05a9ffffa650b953c3217b39b84aef72add04a5d490eee9e7f194a290a",
    "tools/sync/wiki_git.py": "adaea932acfea3829bbc1f12ea6fbf59948d8ebee530f6351b013ad68d78d1ce"
  },
  "index_version": {
    "version": "904884c1a86a45f39bbeccaec0e5e823",
    "git_revision": "c3ec9f611e645a2fe14fe66d518eb0924bf9bb40",
    "created_at_utc": "2026-11-08T20:49:39.997566+00:00",
    "source_sha256": "201798bf276106986e7ef01b4d6357960cb495d09c9855b7df922c1df4fb35a7",
    "documents_sha256": "8b7eea54f6233cdc9d16610296d32b9336ab0cbaa4b881872a088c05464960be",
    "chunks_sha256": "8357d1e0641206990d26827b94f37916661595261f368e30f828bccd03171c95",
    "embedding_model": "intfloat/multilingual-e5-small",
    "publication_policy": "explicit-false-hidden-v1",
    "embedding_fingerprint": "d066fb9a266a1a0c77c65da75dcf538cfab5c87d5c605ff26b1fd2da8a12aa92"
  },
  "knowledge_version": "33c3062af9f1aee887e8950f3ec51b8f",
  "model": {
    "provider": "openai_compatible",
    "model": "model120-fast",
    "endpoint": "https://ui.tukekemt.xyz/api/chat/completions"
  },
  "temperature": 0,
  "structured_output": "prompt",
  "limit": 10,
  "cases": 12,
  "categories": {
    "audit_regression": 12
  },
  "threshold": 0.94,
  "answer_retries": 0,
  "judge_format_attempts": 2,
  "previous_exact_question_overlap": [
    "D01",
    "D02",
    "D03",
    "D04",
    "D05",
    "D06",
    "D07",
    "D08",
    "D09",
    "D10",
    "D11",
    "D12"
  ],
  "limitations": [
    "Question-held-out on an existing public corpus, not unseen documents or model training data.",
    "Gold authored and checked against raw Markdown by AI; no human gold review claimed.",
    "Same configured model judges answers in a separate call; correlated errors remain possible.",
    "Several questions share source profiles; Wilson intervals are descriptive IID intervals, not a cluster-corrected population guarantee.",
    "No runtime changes or TEST-directed tuning are allowed after this freeze. A rerun is reproduction."
  ]
}