{
  "case_stages": 4,
  "independent_business_cases": 1,
  "synthetic": true,
  "human_participants": 0,
  "real_business_execution": false,
  "external_checks": {
    "status": "completed",
    "passed": 48,
    "total": 48
  },
  "component_comparison": {
    "planned_calls": 4,
    "dispatched_records": 1,
    "completed": 0,
    "failed": 1,
    "status": "partial_or_blocked",
    "human_participants": 0,
    "AI_incremental_business_value": "not_established",
    "frozen_source_matches_current": false,
    "interpretation": "One authored case in two information states; four calls planned, not four independent business samples.",
    "tool_boundary": "Both models receive precomputed results. No interactive Python or tool selection; not a general tool-enabled-assistant benchmark.",
    "trials": [
      {
        "group": "G",
        "case_id": "F01",
        "status": "failed",
        "diagnostic": {
          "category": "nonzero_exit_or_missing_output",
          "elapsed_seconds": 0.151,
          "returncode": 1,
          "event_error_count": 0,
          "usage": null
        }
      }
    ]
  },
  "source_sha256": {
    "__init__.py": "5e567e37c2edcc5bbdb510bd5785b55c6ac0bcdf0411a6b0ba9bd790d7f7e627",
    "__main__.py": "3129adff95e0f38b87ab34cddfc27082c7892bcaec267ff5d40f160fdf55b44a",
    "cli.py": "3792b4d311a7dc3a9665d1564b3de8863fdd8d74d2c3d95c0a0b7b4f23fe9bc5",
    "codex_adapter.py": "b2b0c18b930e2c88ea2668554aa428a01e0b5222e7a7be1545185801d5df583f",
    "dashboard.py": "e212178009b3dde9b67471db3d386bd776a6ab017b3cb2d01860fbba8ca4aea0",
    "data.py": "a964c97df5bbd4f265e564e340cf11c6d79abae02687f49eca9f1ce146d3cf21",
    "evaluation.py": "6a58177cb88b8adbbf51a22a4b9d07455589e703b3ac7b5e82b1ec606f36f1d7",
    "fresh.py": "13779bdb12f52de1b85c0056a4db79b02f0f798aad734f39f858b068c330b42e",
    "model.py": "8b1b1c6bd33940f64652e053ef15bfba6294bdee1ed4c0fda5c8f06685c7d448",
    "orchestrator.py": "840fe5c9548e985fc843cb412e06b8f7eef4697f780c97437cfd5f208488b0e5",
    "policy.py": "72f823baf1c5eb41b93b38dd7e4f92a54175614f268b3c346fbc500efb5b8dc8",
    "protocol.py": "dfeef9495fc5d52e7b3d4e642111b081ba73dc17ac390686f28ebe6d605cf27a",
    "provenance.py": "bd2161bb05288defb57efb1475a76dd9f1f711c97c5ec8c7183b8f9464b9ce18",
    "sdk_agent.py": "fce5b1effec8feb0975dc58d63b5db200c3718cfe5dad3e1b3aee1bfe91708e9",
    "tools.py": "51c560ea5f219d93e9ea09158fb477abeb84a3113bcc93f06f52211dbffe1393"
  },
  "decisions": [
    {
      "case": "F01",
      "decision": "INVESTIGATE",
      "action": "fresh_evidence_review"
    },
    {
      "case": "F02",
      "decision": "ESCALATE",
      "action": "fresh_commercial_review"
    },
    {
      "case": "F03",
      "decision": "INVESTIGATE",
      "action": "fresh_reversal_review"
    },
    {
      "case": "F04",
      "decision": "ESCALATE",
      "action": "fresh_source_reconciliation"
    }
  ]
}