{
  "schema_version": 1,
  "evidence_kind": "new actual-member experiment; not historical replay",
  "date": "2026-09-20",
  "run_id": "matrix-66d4dcef85",
  "verdict": "FAIL",
  "original_failures": [
    {
      "check": "matrix_incomplete"
    },
    {
      "case": "reflection",
      "check": "case_failed"
    },
    {
      "case": "multi_level_hierarchy",
      "check": "case_failed"
    },
    {
      "case": "ensemble_judge",
      "check": "case_failed"
    },
    {
      "check": "provenance",
      "error_type": "OSError"
    },
    {
      "case": "ensemble_synthesizer",
      "check": "case_failed"
    },
    {
      "check": "provenance",
      "error_type": "OSError"
    },
    {
      "case": "ensemble_vote",
      "check": "case_failed"
    },
    {
      "check": "provenance",
      "error_type": "OSError"
    },
    {
      "check": "accounting",
      "error_type": "OperationalError"
    }
  ],
  "victor_commit": "307a52fca8d4180c371a1fe6613baed8872599ca",
  "source_clean_at_launch": true,
  "harness_sha256": "d6e3f8f7f6d46ff3918c9e7d23fec88eb35f0c6ce4713af79189357c6de09b41",
  "provider": "zai",
  "model": "glm-5.3",
  "case_contract_version": 1,
  "cases": [
    {
      "case": "sequential",
      "passed": true,
      "failures": [],
      "elapsed_seconds": 51.352,
      "calls": 6,
      "member_usage_checks_passed": 2,
      "member_usage_checks": 2,
      "missing_artifacts": {},
      "independent_pytest_exit_codes": [
        0,
        0
      ],
      "private_report_sha256": "9adf21be2beea54d025e9bd06d1f012fe509b148a1f594e35ff160edab1effff"
    },
    {
      "case": "parallel",
      "passed": true,
      "failures": [],
      "elapsed_seconds": 19.311,
      "calls": 6,
      "member_usage_checks_passed": 2,
      "member_usage_checks": 2,
      "missing_artifacts": {},
      "independent_pytest_exit_codes": [
        0,
        0
      ],
      "private_report_sha256": "0829e0c813cb880c6c2f6cdc3ba9e8c8bb7850ebc732fb216b96c69ca5709a9b"
    },
    {
      "case": "hierarchical",
      "passed": true,
      "failures": [],
      "elapsed_seconds": 32.268,
      "calls": 6,
      "member_usage_checks_passed": 2,
      "member_usage_checks": 2,
      "missing_artifacts": {},
      "independent_pytest_exit_codes": [
        0,
        0
      ],
      "private_report_sha256": "67a58159528d212bdea567731bcaa2edb435e0e92ff869c44de6792ed438e3b6"
    },
    {
      "case": "pipeline",
      "passed": true,
      "failures": [],
      "elapsed_seconds": 43.353,
      "calls": 8,
      "member_usage_checks_passed": 2,
      "member_usage_checks": 2,
      "missing_artifacts": {},
      "independent_pytest_exit_codes": [
        0,
        0
      ],
      "private_report_sha256": "e00518275ebfc0eeaa0adcb3e77357fc656058608155aa123d12c365f2ecf4cf"
    },
    {
      "case": "consensus",
      "passed": true,
      "failures": [],
      "elapsed_seconds": 17.702,
      "calls": 6,
      "member_usage_checks_passed": 2,
      "member_usage_checks": 2,
      "missing_artifacts": {},
      "independent_pytest_exit_codes": [
        0,
        0
      ],
      "private_report_sha256": "95b9cd4f737c0f357dca76acb736f40cb54aecd1c63625f9168d781e51b29c97"
    },
    {
      "case": "reflection",
      "passed": false,
      "failures": [
        "missing_artifacts:second",
        "formation_contract:FileNotFoundError"
      ],
      "elapsed_seconds": 34.577,
      "calls": 7,
      "member_usage_checks_passed": 2,
      "member_usage_checks": 2,
      "missing_artifacts": {
        "second": [
          "review.json"
        ]
      },
      "independent_pytest_exit_codes": [
        0
      ],
      "private_report_sha256": "975a074f9cd81f37b146dd0ffb5ad40658e87c6155302ce7c2cac77d763a0bb7"
    },
    {
      "case": "group_chat",
      "passed": true,
      "failures": [],
      "elapsed_seconds": 30.747,
      "calls": 6,
      "member_usage_checks_passed": 2,
      "member_usage_checks": 2,
      "missing_artifacts": {},
      "independent_pytest_exit_codes": [
        0,
        0
      ],
      "private_report_sha256": "35ffa4624d1b5075d9e6381983dc938408b8b90b03f758ffdd88720f624fce32"
    },
    {
      "case": "debate",
      "passed": true,
      "failures": [],
      "elapsed_seconds": 50.926,
      "calls": 10,
      "member_usage_checks_passed": 3,
      "member_usage_checks": 3,
      "missing_artifacts": {},
      "independent_pytest_exit_codes": [
        0,
        0
      ],
      "private_report_sha256": "1379221cefc959170249fb1f4807fc9e699ff533086d8ca4a1e3902d18249640"
    },
    {
      "case": "handoff",
      "passed": true,
      "failures": [],
      "elapsed_seconds": 26.017,
      "calls": 6,
      "member_usage_checks_passed": 2,
      "member_usage_checks": 2,
      "missing_artifacts": {},
      "independent_pytest_exit_codes": [
        0,
        0
      ],
      "private_report_sha256": "4ae3653ddc42ee64bb16c499847d51c7de28a42614c48d24626389dd53fa6da4"
    },
    {
      "case": "adaptive",
      "passed": true,
      "failures": [],
      "elapsed_seconds": 24.11,
      "calls": 6,
      "member_usage_checks_passed": 2,
      "member_usage_checks": 2,
      "missing_artifacts": {},
      "independent_pytest_exit_codes": [
        0,
        0
      ],
      "private_report_sha256": "8ec90fbabdbda28d5a742fea44f6c91310b786b3e7097b89b7af2e411247fe1e"
    },
    {
      "case": "dynamic_router",
      "passed": true,
      "failures": [],
      "elapsed_seconds": 14.002,
      "calls": 3,
      "member_usage_checks_passed": 1,
      "member_usage_checks": 1,
      "missing_artifacts": {},
      "independent_pytest_exit_codes": [
        0
      ],
      "private_report_sha256": "f1334eb53423da3b874441f7449c4fa7115b11bbce088970cb1beca369c9c862"
    },
    {
      "case": "multi_level_hierarchy",
      "passed": false,
      "failures": [
        "missing_artifacts:second",
        "pytest:second",
        "missing_artifacts:third",
        "pytest:third"
      ],
      "elapsed_seconds": 61.928,
      "calls": 13,
      "member_usage_checks_passed": 3,
      "member_usage_checks": 3,
      "missing_artifacts": {
        "second": [
          "second.py",
          "test_second.py"
        ],
        "third": [
          "third.py",
          "test_third.py"
        ]
      },
      "independent_pytest_exit_codes": [
        0,
        4,
        4
      ],
      "private_report_sha256": "fa0e41e1e9c595392a0c8b7f138f810f10e7900940d75a1e1c77c45f3e9c7a66"
    },
    {
      "case": "ensemble_judge",
      "passed": false,
      "failures": [
        "member_outcomes",
        "distinct_member_sessions",
        "wire_routes_or_attempts",
        "missing_artifacts:first",
        "pytest:first",
        "missing_artifacts:second",
        "pytest:second",
        "missing_worktree:judge",
        "member_usage:executor_2fc125da",
        "member_usage:executor_815f9f99",
        "ensemble_contract",
        "decision_contract:KeyError",
        {
          "check": "provenance",
          "error_type": "OSError"
        }
      ],
      "elapsed_seconds": 7.253,
      "calls": 0,
      "member_usage_checks_passed": 0,
      "member_usage_checks": 2,
      "missing_artifacts": {
        "first": [
          "member.py",
          "test_member.py"
        ],
        "second": [
          "member.py",
          "test_member.py"
        ]
      },
      "independent_pytest_exit_codes": [
        null,
        null
      ],
      "private_report_sha256": "2126dc05054ed7011d494397e8e0d2de088c4a6b39efa7a1f80a4f96f5483f6a"
    },
    {
      "case": "ensemble_synthesizer",
      "passed": false,
      "failures": [
        "execution:OSError",
        {
          "check": "provenance",
          "error_type": "OSError"
        }
      ],
      "elapsed_seconds": 0.0,
      "calls": 0,
      "member_usage_checks_passed": 0,
      "member_usage_checks": 0,
      "missing_artifacts": {},
      "independent_pytest_exit_codes": [],
      "private_report_sha256": "9d07ce525ab64e97fd7d2fc39e8fa219e0fed6bc97f7e52c250c97820df3ee02"
    },
    {
      "case": "ensemble_vote",
      "passed": false,
      "failures": [
        "execution:OSError",
        {
          "check": "provenance",
          "error_type": "OSError"
        }
      ],
      "elapsed_seconds": 0.0,
      "calls": 0,
      "member_usage_checks_passed": 0,
      "member_usage_checks": 0,
      "missing_artifacts": {},
      "independent_pytest_exit_codes": [],
      "private_report_sha256": "e178f63b1c3e6e59a732707dc2b902c6b241a3940acf0764c68af161ac380039"
    }
  ],
  "progress": {
    "case_passes": 10,
    "planned_zai_cases": 15,
    "deliverable_failures": 2,
    "resource_blocked_ensemble_cases": 3,
    "new_matched_qwen_cases_run": 0,
    "planned_matched_cases": 30
  },
  "supplemental_accounting": {
    "verdict": "PASS",
    "new_model_calls": 0,
    "original_verdict_changed": false,
    "original_before_snapshot_verified_at_boundary": true,
    "calls": 83,
    "distinct_sessions": 25,
    "totals": {
      "tokens_in": 37280,
      "tokens_out": 12594,
      "cache_read_tokens": 212608,
      "cache_creation_tokens": 0
    },
    "inclusive_input_tokens": 249888,
    "total_tokens": 262482,
    "unrelated_new_rows": 0,
    "failures": [],
    "report_sha256": "28912dc5673d1da4be4669f24c21317d630cfdbf98179b17212135fdbecffdf8"
  },
  "sandhi": {
    "source_commit": "eb38ff4b4c90121ac3f2e2e033ac5d5b2d057e2d",
    "binary_sha256": "5933e8c702d482d075d9b88a579a8bb0d6feaa50b8e8e2ed69ccd5ad23c23826",
    "pid": 47910,
    "gateway_url": "http://127.0.0.1:18788",
    "inferflux_upstream": "http://127.0.0.1:18081/v1",
    "rollback_upstream": "http://127.0.0.1:18080/v1"
  },
  "timeouts": {
    "buffered_gateway_seconds": 120,
    "case_seconds": 240,
    "observed_http_errors": 0,
    "observed_deadline_failures": 0,
    "max_request_seconds": 7.535
  },
  "limitations": [
    "Original overall verdict remains FAIL; resource exhaustion prevented its final accounting. Supplemental reconciliation is a separate read-only result.",
    "ZAI case passes do not establish matched Qwen or full mixed-team C5 acceptance.",
    "No inference calls were repeated to obtain this evidence summary or supplemental reconciliation.",
    "Request/response bodies and credentials are not exported.",
    "Buffered calls only; no streaming, cancellation, tokenizer-unit, executed-cache-reuse or session-lease acceptance.",
    "File-descriptor exhaustion root cause is not established by this run.",
    "The separately ready InferFlux runtime was partial CUDA offload, not R9700/full-GPU acceptance; no InferFlux calls belong to this experiment.",
    "Historical failed reports and passing five-call streaming evidence remain unchanged."
  ]
}