{
  "lab_id": "LAB-230-01",
  "recorded_at": "2026-09-07T14:47:48.183851+00:00",
  "environment": "Python 3.14.4",
  "source_sha256": "55c96fd2d12b02f70480cc72e044a35dd12a7d0b7a7f947ce4f8b1af216dbda4",
  "validation_output": "test_every_citation_resolves_to_an_active_exact_version (test_citations.CitationTests.test_every_citation_resolves_to_an_active_exact_version) ... ok\ntest_deleted_document_is_absent_from_index_context_and_citations (test_no_answer.NoAnswerAndDeletionTests.test_deleted_document_is_absent_from_index_context_and_citations) ... ok\ntest_unsupported_question_uses_approved_no_answer (test_no_answer.NoAnswerAndDeletionTests.test_unsupported_question_uses_approved_no_answer) ... ok\ntest_denied_finance_document_never_enters_hr_context (test_permissions.PermissionTests.test_denied_finance_document_never_enters_hr_context) ... ok\ntest_permitted_finance_user_receives_current_version (test_permissions.PermissionTests.test_permitted_finance_user_receives_current_version) ... ok\ntest_retrieved_instruction_is_detected_and_removed (test_prompt_injection.PromptInjectionTests.test_retrieved_instruction_is_detected_and_removed) ... ok\n\n----------------------------------------------------------------------\nRan 6 tests in 0.005s\n\nOK\n",
  "sample": {
    "question": "finance budget approval threshold",
    "allowed": {
      "answer": "Finance budget requests above 50000 credits require CFO approval and a recorded risk review.",
      "citations": [
        {
          "source_id": "FIN-BUDGET-2026",
          "version": "3.0"
        }
      ],
      "context": [
        {
          "source_id": "FIN-BUDGET-2026",
          "version": "3.0",
          "content": "Finance budget requests above 50000 credits require CFO approval and a recorded risk review.",
          "allowed_groups": [
            "finance"
          ]
        }
      ],
      "evidence": {
        "schema_version": "1.0.0",
        "recorded_at": "2026-09-07T14:47:48.163678+00:00",
        "lab_id": "LAB-230-01",
        "user_id": "fiona.finance",
        "question_hash": "edf7475724ccf5cbc909bd467361da42635bb0193e08f8fdec76966f2ca9a1bd",
        "versions": {
          "corpus": "f3fc499d347b",
          "index": "97d746eed4b2",
          "retriever": "stdlib-token-overlap-0.1.0",
          "permission_policy": "1.0.0",
          "generation_model": "deterministic-context-assembler-0.1.0",
          "prompt": "governing-prompt-0.1.0",
          "evaluation_dataset": "synthetic-questions-0.1.0"
        },
        "controls": {
          "permission_filter_before_scoring": true,
          "context_source_count": 1,
          "injection_detected": false,
          "all_citations_resolve": true
        },
        "citations": [
          {
            "source_id": "FIN-BUDGET-2026",
            "version": "3.0"
          }
        ],
        "known_limitations": [
          "Synthetic corpus and identities only",
          "Lexical retrieval is a control demonstration, not a production relevance benchmark",
          "No production identity provider or language model is included"
        ]
      }
    },
    "denied": {
      "answer": "No permitted source supports an answer to this question.",
      "citations": [],
      "context": [],
      "evidence": {
        "schema_version": "1.0.0",
        "recorded_at": "2026-09-07T14:47:48.163739+00:00",
        "lab_id": "LAB-230-01",
        "user_id": "harper.hr",
        "question_hash": "edf7475724ccf5cbc909bd467361da42635bb0193e08f8fdec76966f2ca9a1bd",
        "versions": {
          "corpus": "f3fc499d347b",
          "index": "97d746eed4b2",
          "retriever": "stdlib-token-overlap-0.1.0",
          "permission_policy": "1.0.0",
          "generation_model": "deterministic-context-assembler-0.1.0",
          "prompt": "governing-prompt-0.1.0",
          "evaluation_dataset": "synthetic-questions-0.1.0"
        },
        "controls": {
          "permission_filter_before_scoring": true,
          "context_source_count": 0,
          "injection_detected": false,
          "all_citations_resolve": true
        },
        "citations": [],
        "known_limitations": [
          "Synthetic corpus and identities only",
          "Lexical retrieval is a control demonstration, not a production relevance benchmark",
          "No production identity provider or language model is included"
        ]
      }
    }
  },
  "scope": "Synthetic deterministic control prototype; not a production benchmark or independent validation."
}