{
  "schema_version": 1,
  "updated_at": "2026-11-17T21:16:35.130Z",
  "source_revision": "c0deccb99807b2539df726e4b09091e12ce483e32d26f3952f29852a3ca14c20",
  "source_revision_kind": "sha256_of_checked_source_files",
  "contract": {
    "status": "passed",
    "checks": 101,
    "tasks": 8,
    "scope": "62 strict-search contract checks and 39 authored verifier-calibration cases. These are deterministic checks, not model trials or image-quality certification.",
    "groups": [
      {
        "name": "strict_search_requirements",
        "checks": 62,
        "kind": "deterministic_service_contract"
      },
      {
        "name": "outcome_verifier_calibration",
        "checks": 39,
        "kind": "deterministic_verifier_calibration",
        "known_good_cases": 16,
        "negative_cases": 23
      }
    ]
  },
  "agent_trials": {
    "status": "completed",
    "trials": 12,
    "passed": 12,
    "mode": "Cooperative synthetic fixture diagnostics",
    "model": "Session agent (exact model ID not recorded)",
    "cost_usd": null,
    "observed_at": "2026-11-16T22:36:14.138Z",
    "evaluated_interface_current": true,
    "stale_files": [],
    "variants": [
      {
        "variant": "baseline",
        "models": [
          "session-agent"
        ],
        "harnesses": [
          "fresh-collaboration-agent"
        ],
        "service_module_sha256": [
          "fb3f16c8c8525f87c3d0434b45398d2df8c6dc36e76d403e7333f71e6f133755"
        ],
        "trial_count": 6,
        "passed": 6,
        "per_task": [
          {
            "task_id": "native-image",
            "status": "run",
            "trials": 2,
            "passed": 2,
            "success_rate": 1,
            "all_observed_trials_passed": true,
            "average_tool_calls_successful": 1,
            "average_duration_ms_successful": 7363,
            "total_cost_usd": null,
            "cost_per_success_including_failures": null
          },
          {
            "task_id": "exact-variant",
            "status": "run",
            "trials": 2,
            "passed": 2,
            "success_rate": 1,
            "all_observed_trials_passed": true,
            "average_tool_calls_successful": 2,
            "average_duration_ms_successful": 12560,
            "total_cost_usd": null,
            "cost_per_success_including_failures": null
          },
          {
            "task_id": "numeric-variant",
            "status": "not_run",
            "trials": 0,
            "passed": 0,
            "success_rate": null,
            "all_observed_trials_passed": null,
            "average_tool_calls_successful": null,
            "average_duration_ms_successful": null,
            "total_cost_usd": null,
            "cost_per_success_including_failures": null
          },
          {
            "task_id": "ambiguous-identifier",
            "status": "not_run",
            "trials": 0,
            "passed": 0,
            "success_rate": null,
            "all_observed_trials_passed": null,
            "average_tool_calls_successful": null,
            "average_duration_ms_successful": null,
            "total_cost_usd": null,
            "cost_per_success_including_failures": null
          },
          {
            "task_id": "impossible-native-size",
            "status": "run",
            "trials": 2,
            "passed": 2,
            "success_rate": 1,
            "all_observed_trials_passed": true,
            "average_tool_calls_successful": 1,
            "average_duration_ms_successful": 7633,
            "total_cost_usd": null,
            "cost_per_success_including_failures": null
          },
          {
            "task_id": "unknown-facts",
            "status": "not_run",
            "trials": 0,
            "passed": 0,
            "success_rate": null,
            "all_observed_trials_passed": null,
            "average_tool_calls_successful": null,
            "average_duration_ms_successful": null,
            "total_cost_usd": null,
            "cost_per_success_including_failures": null
          },
          {
            "task_id": "distinct-batch",
            "status": "not_run",
            "trials": 0,
            "passed": 0,
            "success_rate": null,
            "all_observed_trials_passed": null,
            "average_tool_calls_successful": null,
            "average_duration_ms_successful": null,
            "total_cost_usd": null,
            "cost_per_success_including_failures": null
          },
          {
            "task_id": "protected-original",
            "status": "not_run",
            "trials": 0,
            "passed": 0,
            "success_rate": null,
            "all_observed_trials_passed": null,
            "average_tool_calls_successful": null,
            "average_duration_ms_successful": null,
            "total_cost_usd": null,
            "cost_per_success_including_failures": null
          }
        ]
      },
      {
        "variant": "candidate",
        "models": [
          "session-agent"
        ],
        "harnesses": [
          "fresh-collaboration-agent"
        ],
        "service_module_sha256": [
          "1edb26c4bb5f25737ea7e57ea252b9dfecf960167049467569b55766cdd095ac"
        ],
        "trial_count": 6,
        "passed": 6,
        "per_task": [
          {
            "task_id": "native-image",
            "status": "run",
            "trials": 2,
            "passed": 2,
            "success_rate": 1,
            "all_observed_trials_passed": true,
            "average_tool_calls_successful": 1,
            "average_duration_ms_successful": 8944.5,
            "total_cost_usd": null,
            "cost_per_success_including_failures": null
          },
          {
            "task_id": "exact-variant",
            "status": "run",
            "trials": 2,
            "passed": 2,
            "success_rate": 1,
            "all_observed_trials_passed": true,
            "average_tool_calls_successful": 1,
            "average_duration_ms_successful": 13812.5,
            "total_cost_usd": null,
            "cost_per_success_including_failures": null
          },
          {
            "task_id": "numeric-variant",
            "status": "not_run",
            "trials": 0,
            "passed": 0,
            "success_rate": null,
            "all_observed_trials_passed": null,
            "average_tool_calls_successful": null,
            "average_duration_ms_successful": null,
            "total_cost_usd": null,
            "cost_per_success_including_failures": null
          },
          {
            "task_id": "ambiguous-identifier",
            "status": "not_run",
            "trials": 0,
            "passed": 0,
            "success_rate": null,
            "all_observed_trials_passed": null,
            "average_tool_calls_successful": null,
            "average_duration_ms_successful": null,
            "total_cost_usd": null,
            "cost_per_success_including_failures": null
          },
          {
            "task_id": "impossible-native-size",
            "status": "run",
            "trials": 2,
            "passed": 2,
            "success_rate": 1,
            "all_observed_trials_passed": true,
            "average_tool_calls_successful": 1,
            "average_duration_ms_successful": 8174.5,
            "total_cost_usd": null,
            "cost_per_success_including_failures": null
          },
          {
            "task_id": "unknown-facts",
            "status": "not_run",
            "trials": 0,
            "passed": 0,
            "success_rate": null,
            "all_observed_trials_passed": null,
            "average_tool_calls_successful": null,
            "average_duration_ms_successful": null,
            "total_cost_usd": null,
            "cost_per_success_including_failures": null
          },
          {
            "task_id": "distinct-batch",
            "status": "not_run",
            "trials": 0,
            "passed": 0,
            "success_rate": null,
            "all_observed_trials_passed": null,
            "average_tool_calls_successful": null,
            "average_duration_ms_successful": null,
            "total_cost_usd": null,
            "cost_per_success_including_failures": null
          },
          {
            "task_id": "protected-original",
            "status": "not_run",
            "trials": 0,
            "passed": 0,
            "success_rate": null,
            "all_observed_trials_passed": null,
            "average_tool_calls_successful": null,
            "average_duration_ms_successful": null,
            "total_cost_usd": null,
            "cost_per_success_including_failures": null
          }
        ]
      }
    ],
    "reliability": {
      "status": "not_estimated",
      "pass_at_k": null,
      "pass_power_k": null,
      "reason": "No probabilistic reliability claim is made from cooperative fixture smoke trials; independent repeated production trials are a separate evaluation."
    },
    "historical_report": {
      "observed_at": "2026-11-16T22:36:14.138Z",
      "trials": 12,
      "sha256": "c22ebdff7eaf7b8155b46334d8a2dccf5d40293ab284181afbb6e96e3b4071dc",
      "current": true
    }
  },
  "provenance": {
    "source_files": [
      {
        "path": "src/image-data-service.mjs",
        "sha256": "1edb26c4bb5f25737ea7e57ea252b9dfecf960167049467569b55766cdd095ac"
      },
      {
        "path": "src/catalog-search.mjs",
        "sha256": "5aa1e6c512c2b3beefff378ea257828952943763ccb3a78496c04f1aaea3a91c"
      },
      {
        "path": "src/worker-runtime.mjs",
        "sha256": "ec615634ec66a250a76bdf033a1bf673d6d4ca7e7cfc1603f1c3fc215a9308b9"
      },
      {
        "path": "src/developer-runtime.mjs",
        "sha256": "8c21c553ff915e8b78dd1f6360064463dd604b5f17d5ba91177434808815ae49"
      },
      {
        "path": "src/agent-entry.mjs",
        "sha256": "50236948445de3c3b3a65d2e6c8c9ea93600e23b53d923b356f0eacac24d76e8"
      },
      {
        "path": "scripts/check-image-requirements.mjs",
        "sha256": "74835ca91a883f9fe5d360dc178f3f13cd8e3daac69bc8752a6852797c25b034"
      },
      {
        "path": "scripts/check-agent-eval-verifier.mjs",
        "sha256": "81d5752e541dcc984260f9fa2744a86cde2d1bad337f5aad25c78c9286c53bdf"
      },
      {
        "path": "scripts/prepare-agent-evidence.mjs",
        "sha256": "d6be2295b0f22585fbb9e27882f3d76d03b35cd8212d8b2493ba953bf82d9ad0"
      },
      {
        "path": "scripts/check-agent-entry.mjs",
        "sha256": "06fdffa1ab41e8a4702e10343a7f465546799374f3de2e65088efbf33cab98bf"
      },
      {
        "path": "evals/product-images/taskset.json",
        "sha256": "4e1ce24babbd8a0564b3176f422aa1f350bc649b7714a9eaf654053248665ce3"
      },
      {
        "path": "evals/product-images/fixtures/source-records.json",
        "sha256": "5e507bfdebcfeb4646d395f6666e7e261a09f6b1fdef79704baf7580af1feba4"
      },
      {
        "path": "evals/product-images/fixtures.mjs",
        "sha256": "5cbe0319f8bbf7e0872b1f83d6a220a1de98bd8e918611670f6b222617b7d381"
      },
      {
        "path": "evals/product-images/fixture-service.mjs",
        "sha256": "f7d2b8022ed15e4abaed2201912e1f54950509767e6111814856446c0b5f3421"
      },
      {
        "path": "evals/product-images/verify.mjs",
        "sha256": "e135d1700cfc062d3b2eb0e88e833d679ea74637e6566081be581136b984122c"
      },
      {
        "path": "evals/product-images/references.mjs",
        "sha256": "0f56fd416c9e02255f75b6856ef83a82893190be1b06c4f43ab1554cfc9169d0"
      }
    ],
    "evaluated_modules": [
      {
        "path": "src/image-data-service.mjs",
        "sha256": "1edb26c4bb5f25737ea7e57ea252b9dfecf960167049467569b55766cdd095ac"
      },
      {
        "path": "evals/product-images/baselines/v144-image-data-service.mjs",
        "sha256": "fb3f16c8c8525f87c3d0434b45398d2df8c6dc36e76d403e7333f71e6f133755"
      },
      {
        "path": "evals/product-images/taskset.json",
        "sha256": "4e1ce24babbd8a0564b3176f422aa1f350bc649b7714a9eaf654053248665ce3"
      },
      {
        "path": "evals/product-images/fixtures/source-records.json",
        "sha256": "5e507bfdebcfeb4646d395f6666e7e261a09f6b1fdef79704baf7580af1feba4"
      },
      {
        "path": "evals/product-images/fixtures.mjs",
        "sha256": "5cbe0319f8bbf7e0872b1f83d6a220a1de98bd8e918611670f6b222617b7d381"
      },
      {
        "path": "evals/product-images/fixture-service.mjs",
        "sha256": "f7d2b8022ed15e4abaed2201912e1f54950509767e6111814856446c0b5f3421"
      },
      {
        "path": "evals/product-images/verify.mjs",
        "sha256": "e135d1700cfc062d3b2eb0e88e833d679ea74637e6566081be581136b984122c"
      },
      {
        "path": "scripts/run-agent-evals.mjs",
        "sha256": "781079c4514ba5e78470161a32af29c7864f183baa174dd9d3fc73b13af7f710"
      }
    ],
    "taskset_sha256": "4e1ce24babbd8a0564b3176f422aa1f350bc649b7714a9eaf654053248665ce3",
    "fixture_sha256": "5e507bfdebcfeb4646d395f6666e7e261a09f6b1fdef79704baf7580af1feba4",
    "taskset_revision": "2026-11-16.1",
    "fixture_id": "synthetic-image-records-v1",
    "verifier_version": "product-image-verifier-v1",
    "verifier_sha256": "e135d1700cfc062d3b2eb0e88e833d679ea74637e6566081be581136b984122c",
    "baseline_source_commit": "909cc43a53ea35d322b2a61ddc4c1ce121b3aca6",
    "model_report_sha256": "c22ebdff7eaf7b8155b46334d8a2dccf5d40293ab284181afbb6e96e3b4071dc"
  },
  "limitations": [
    "Fresh agent trials covered three of the eight authored task types, twice per interface. The five remaining types have verifier coverage but no fresh-agent trial result.",
    "Baseline and updated interfaces both passed six of six recorded trials. This sample does not demonstrate a success-rate improvement.",
    "The shared fixture search adapter does not execute production search. A baseline variant trial used a redundant second lookup; fewer observed calls do not establish a causal speed improvement.",
    "Synthetic fixtures exercise constrained decisions, not the production catalog, network delivery, visual product fidelity, search discovery from the open web, paid entitlements or cross-model reliability.",
    "The local collector separates tool traces from final answers but is not a hostile sandbox. Filesystem access could alter fixture or trace files; use external isolation and collector-owned artifacts for adversarial evaluations.",
    "The verifier grades the typed handoff artifact. It does not grade arbitrary conversational prose or unstated claims.",
    "Only successful trials contribute to efficiency averages. Failed attempts remain in success denominators. Missing cost stays null.",
    "A perfect small sample means no failures were observed in those cases, not that the service or an agent always succeeds.",
    "Exact model identity and token cost were not recorded. Timing starts at first CLI use, excludes queue time and is missing for three early baseline trials; this is not a production-latency comparison."
  ],
  "methodology_url": "/agents.md"
}
