{
  "reviewedAt": "2026-10-06",
  "verification": "读取已有原始报告；本轮未重新运行底层评测",
  "acceptance": {
    "generated_at": "2026-10-06T01:55:01.209506Z",
    "status": "passed",
    "total": 17,
    "passed": 17,
    "failed": 0,
    "duration_ms": 296.5507000335492
  },
  "generalization": {
    "generated_at": "2026-10-06T01:51:47.260451Z",
    "status": "needs_attention",
    "total": 34,
    "passed": 30,
    "failed": 4,
    "pass_rate": 0.8823529411764706
  },
  "groups": [
    {
      "id": "typical",
      "name": "典型业务场景",
      "total": 6,
      "passed": 6,
      "failed": 0
    },
    {
      "id": "linguistic_generalization",
      "name": "表达变化与语义泛化",
      "total": 7,
      "passed": 3,
      "failed": 4
    },
    {
      "id": "boundaries",
      "name": "输入边界",
      "total": 7,
      "passed": 7,
      "failed": 0
    },
    {
      "id": "abnormal_and_security",
      "name": "异常输入与安全",
      "total": 9,
      "passed": 9,
      "failed": 0
    },
    {
      "id": "stability",
      "name": "重复性与并发稳定性",
      "total": 5,
      "passed": 5,
      "failed": 0
    }
  ],
  "limits": [
    "The default hashing-v1 embedding is lexical and is not a production semantic model.",
    "The deterministic mock generator validates orchestration, not real-model answer quality.",
    "The suite runs on SQLite in one service instance; distributed consistency is not covered.",
    "External model, SearXNG, container, and network failure modes require environment tests."
  ],
  "sources": [
    {
      "file": "outputs/final-acceptance.json",
      "sha256": "def6a8c62b4023eb9b65426f7dcf787aba17529b607838529ea0b316086fa02e",
      "bytes": 6130
    },
    {
      "file": "outputs/generalization-stability-report.json",
      "sha256": "afbe49014da61ce61e737a984b5cc7b9d20b9d9f66d11bceee5c133fcee801cc",
      "bytes": 52470
    }
  ]
}