I3372 — The registered corpus is run in both arms, its records are committed, and its verdict is published with its losses and its origin

✕BLOKOVÁNO vlna 6 · p1 · profil implementation · bezpečné paralelně

Part of a-registered-corpus-judges-majordomus — The registered corpus judges the same agent with and without Majordomus, by an exact task-level test on accepted results, and the verdict is published with its losses.

Cíl

With the owner's recorded consent to the ceiling, dry-run the suite, rehearse one pair to confirm a live run records end to end, then run every declared job in the registered order with the pinned release, commit the run records under .ai/repo/benchmarks/economics/runs/corpus, regenerate the economics reports, and publish the T2 verdict whatever it is, stating that the corpus is one synthetic fixture written by the treatment's builders.

Proč

T2 of the owner's report of 2026-10-19 asks for the comparison to be run; no live run has ever been committed. Under ADR 0082 no run is overwritten or dropped and an unfavourable result is shown.

Aktuální stav

git ls-files .ai/repo/benchmarks/economics/runs lists only runs/context; branch feature/economics-pilot-runs raised pilot repetitions and committed no run.

Cílový stav

runs/corpus holds every attempted run; the T2 row reads a computed status with its losses.

Rozsah

• .ai/repo/benchmarks/economics/runs
• docs/generated/economics.json
• docs/generated/economics.md
• docs/generated/economics.yaml
• site/data/generated/economics.json

Mimo rozsah

• Editing the registration or methodology during the runs
• Re-running or excluding a run to change a verdict
• Changing the calculator; a runner defect becomes its own issue

Závislosti

• I3370✕BLOCKED The corpus suite is registered and committed before its first run, and the T2 protocol names its registration digest
• I3371✕BLOCKED Economics summary decides a registered suite's primary outcome by its registered test, reports secondary differences without a verdict word, and refuses typed figures about them
• I3354✕BLOCKED A pure scorecard function computes each proof test's status, counted sample, effect and losses from committed records, and withholds a verdict until the registered sample is complete

Kritéria přijetí

• The consent decision is recorded before the first live session
• economics run --suite corpus --dry-run lists every job in the registered order
• Every declared pair is attempted, and each missing or invalid pair is named with its reason
• Every run within a pair reports the same harness version, model, fixture digest and pinned release
• The published result states the corpus origin
• majordomus economics check and majordomus generate --check pass

Validace

• majordomus economics summary
• majordomus economics proof
• majordomus economics check
• majordomus generate --check
• bash test/run.sh 431_economics_claims_gated

Požadované důkazy

• corpus_run_records
• corpus_verdict_published
• budget_consent_decision

Důkazy

Nic nezaznamenáno. Za každým tokenem výše musí stát příkaz nebo artefakt, než lze tento úkol dokončit; vyprávění se odmítá.

Časová osa

zahájeno

—

ověřeno

—

dokončeno

—

Tato tři pole, důkazy výše a stav závislostí jsou vše, z čeho se stav skládá. Neexistuje žádné pole stavu, které by s nimi mohlo být v rozporu.

Canonical record: .ai/repo/project/issues/I3372.yaml. Read it back with majordomus plan show I3372.