{
  "schema_version": 1,
  "status": "prepared",
  "verified": true,
  "purpose": "Originality Gate corpus snapshot for run-002",
  "normalization": "Unicode NFC + casefold + whitespace tokenization",
  "ngram_size": 14,
  "thresholds": {
    "ngram14_overlap_max": 8,
    "longest_exact_token_run_max": 30
  },
  "files": [
    {
      "project_id": "001-Ban-Ghi-Am-Cuoi-Cung-Trong-Mua",
      "path": "/data/video-pipeline/HaTramAudio/project/001-Ban-Ghi-Am-Cuoi-Cung-Trong-Mua/story/spoken-narration.txt",
      "size_bytes": 48834,
      "sha256": "911607f6dd37616c31da3b8a830b61f4c4197611855f34dc70547bc019208a9a"
    },
    {
      "project_id": "001-Ban-Ghi-Am-Cuoi-Cung-Trong-Mua",
      "path": "/data/video-pipeline/HaTramAudio/project/001-Ban-Ghi-Am-Cuoi-Cung-Trong-Mua/story/story-canon.txt",
      "size_bytes": 48834,
      "sha256": "911607f6dd37616c31da3b8a830b61f4c4197611855f34dc70547bc019208a9a"
    },
    {
      "project_id": "002-Tang-Hai-Muoi-Bon-Khong-Co-Trong-Ban-Ve",
      "path": "/data/video-pipeline/HaTramAudio/project/002-Tang-Hai-Muoi-Bon-Khong-Co-Trong-Ban-Ve/story/spoken-narration.txt",
      "size_bytes": 73828,
      "sha256": "6280a8c654ba783403280c232716c6bf6eaacb0e021b7902acb463f7a6b3e6d3"
    },
    {
      "project_id": "002-Tang-Hai-Muoi-Bon-Khong-Co-Trong-Ban-Ve",
      "path": "/data/video-pipeline/HaTramAudio/project/002-Tang-Hai-Muoi-Bon-Khong-Co-Trong-Ban-Ve/story/story-canon.txt",
      "size_bytes": 73881,
      "sha256": "8526764f2ccf349aa40f97c8fceb92c94b42e46e50c1e18dbe75181638fb4bbd"
    }
  ],
  "file_count": 4,
  "created_at": "2026-07-19T04:14:38.762939Z"
}