{
  "demo_id": "agent-observability",
  "mode": "artifact_replay",
  "source_repository": "https://github.com/prasadnitish/AI-Eval-Control-Tower/tree/codex/agent-observability",
  "source_commit": "24d5d87fcd9c57ea957bed45c6bc38ed4a25b5e7",
  "generated_at": "2026-07-12T17:44:17.969Z",
  "generator_command": "npm test && npm run trace:failure && npm run trace:load",
  "inputs": [
    {
      "name": "forced-failure trace",
      "path": "forced-failure.json",
      "sha256": "101396cc31fe3962b21fe0559efe0ee840c116dd689879009c32519c83d0ab76",
      "bytes": 1088
    },
    {
      "name": "150,000-span load report",
      "path": "trace-load.json",
      "sha256": "e9268261b6fddc5e13361de6ef97f3b3b0b9235124d82c01930ab1411083888e",
      "bytes": 369
    }
  ],
  "artifacts": [
    {
      "name": "forced-failure trace",
      "path": "forced-failure.json",
      "sha256": "101396cc31fe3962b21fe0559efe0ee840c116dd689879009c32519c83d0ab76",
      "bytes": 1088,
      "media_type": "application/json"
    },
    {
      "name": "150,000-span load report",
      "path": "trace-load.json",
      "sha256": "e9268261b6fddc5e13361de6ef97f3b3b0b9235124d82c01930ab1411083888e",
      "bytes": 369,
      "media_type": "application/json"
    }
  ],
  "environment": {
    "execution": "committed artifact replay with browser-local import",
    "provider_calls": false,
    "timestamp_offsets": "reconstructed from source trace timing"
  },
  "providers": [],
  "claims": [
    {
      "id": "trace-forced-failure",
      "statement": "The committed trace retains the policy-service failure and the release-gate fallback reason.",
      "evidence_artifact": "forced-failure.json",
      "evidence_pointer": "/spans/1/detail/error_code"
    },
    {
      "id": "trace-load",
      "statement": "The in-process collector simulation attempted 150,000 spans and retained all 150 error spans.",
      "evidence_artifact": "trace-load.json",
      "evidence_pointer": "/retained_errors"
    }
  ],
  "limitations": [
    "Load evidence is an in-process buffered collector simulation, not a hosted throughput benchmark.",
    "The fixture contains relative millisecond offsets rather than original wall-clock span timestamps."
  ]
}