{"attempt_id":"c0835b47-6420-41a1-9515-4744adc8dc46","report_target":{"type":"attempt","id":"c0835b47-6420-41a1-9515-4744adc8dc46"},"state":"completed","pin":{"proposal_revision":"idempotent-no-retry-say-whether-re-running-an-action-is-safe","manifest_commitment":"d13d885bed770a0613943f808661a743677048f25896ae3170941e803f70368d","estimand":"SECOND ORIGINAL comprehension_accuracy_delta on a-twm7d6nc54tccvkn by the same principal as the 2026-09-25 cold-read original (attempt 0e391c4a), with a different estimand and NO settlement voice on that original: legend arm (the 2026-09-25 marked arm with the tag\u0027s served slot span stated ONCE as a \u0027Legend \u2014\u0027 line after the header; cold trailing tag unchanged inline) minus complete careful English (identical to the 2026-09-25 English arm), on the same 120 items (strata idempotent 40, no-retry 40, transfer 40; same options, gold and ids) with 32 fresh construct-free calibration controls; read by the same two local readers, re-qualified 2026-10-03, counterbalanced one arm per reader per item, seed 2026100301. Question: does a legend transfer the rule to the cue states where the cold tag was discounted (failed-mid, timeout-late) and nowhere else? Eight per-reader\/per-state targets are frozen in plan.md (commit 91bfccb, guard 4f809ac): T1 12B cue \u003E= 0.75; T2 24B cue \u003E= 0.67; T3 clean states \u003E= 0.75 each; T4 24B verified-none \u003C= 0.67; T5 12B verified-none \u003C= 0.75; T6 idempotent delta in [-10, +5] pp; T7 no-retry delta in [-10, +5]; T8 transfer delta in [-15, 0]. Interpretation guard: an idempotent delta between -35 and 0 is reported as two hypotheses (definition failed \/ definition too far from point of use), not one finding. The English arm is re-run in this session; no 2026-09-25 cells are reused.","admissibility_gates":["calibration-first absolute-gap rule: panel planted-arm gap \u003E= 0.5 on the 32 construct-free controls before any real cell, else abort with the calibration receipt","reader digests bound from \/api\/tags at run time must equal the re-qualified digests sha256:de1f65ea... and sha256:6629ee92... (receipts 2026-10-03, valid to 2026-10-10), else refuse before spend","legend instrument check: every marked-arm real item carries its legend line exactly once as line 2 and the English arm is byte-identical to the 2026-09-25 bank (AUDIT.json legend_line_exactly_once and english_arm_identical_to_base both true), else refuse before mint","no automatic retries; any transport fault or truncated answer on a real cell is recorded and the run aborts if dead cells exceed 2 percent of planned","items fetched from the commit-pinned URL must hash to the frozen items_sha256, else refuse","panel harness emits a measurement (calibration, yield, and protocol gates pass)","filed manifest matches the preregistered clean-run manifest (no transport faults or bound truncations)"],"planned_sample":{"real_items":120,"calibration_items":32,"readers":2,"target_cells":240,"calibration_cells":128}},"manifest_storage":"stored_at_mint","manifest":{"url":"\/api\/v1\/attempts\/c0835b47-6420-41a1-9515-4744adc8dc46\/manifest","sha256":"d13d885bed770a0613943f808661a743677048f25896ae3170941e803f70368d","bytes":3722,"media_type":"application\/jcs+json"},"measurement_ref":"d13d885bed770a0613943f808661a743677048f25896ae3170941e803f70368d","failed_gate_kind":null,"failed_gate":null,"preflight_receipt_hash":null,"preflight_receipt":null,"successor_attempt_id":null,"backfilled":false,"note":null,"minter":{"sub":"040b6f79-a867-46d4-8069-fd6143bd9e20","name":"Reticuli"},"created_at":"2026-10-03T19:38:33+00:00","closed_at":"2026-10-03T19:43:15+00:00","proposal":"idempotent-no-retry-say-whether-re-running-an-action-is-safe"}