I3382 — Every recorded benchmark run answers the same delivery-economics shape as a task

✕BLOKOVÁNO vlna 1 · p1 · profil implementation · bezpečné paralelně

Part of delivery-economics-are-measured-over-declared-windows — Delivery economics are measured on corpus runs and on real work over a window registered before it opens, and published with their denominators and losses.

Cíl

Add economics tasks --runs: per recorded benchmark run, wall time from started_at to finished_at and per session (duration_ms), scripted follow-up prompts apart from interventions (none by construction of the headless harness), ci_reruns not_applicable, spend as observed tokens with the projected cost labelled as a projection, and accepted as the hidden acceptance_tests check. Add an aggregate per suite and variant with medians beside their denominators, published nowhere until ADR 0082's rule is met.

Proč

T3 of the owner's report of 2026-10-19 asks for the same figures on corpus runs, and T2 and T4 compare arms in this shape, so real work and benchmark runs are read alike.

Aktuální stav

apps/majordomus-cli/src/economics/model.rs:687 records duration_ms per session and :701 reported_cost_microusd; no reading combines them per run.

Cílový stav

economics tasks --runs returns one record per committed run in the task shape.

Rozsah

• apps/majordomus-cli/src/economics
• apps/majordomus-cli/tests/economics.rs
• docs/ECONOMICS.md

Mimo rozsah

• Comparing arms (I3371)
• Publishing an aggregate

Závislosti

• I3376○READY One capability states each task's wall time and episodes from every checkout's ledger, and names every field it cannot yet fill as not observed

Kritéria přijetí

• Every committed run under .ai/repo/benchmarks/economics/runs answers the shape
• Interventions read none by construction, ci_reruns not_applicable
• The projected cost is labelled as a projection
• The aggregate states its denominator

Validace

• cargo test --manifest-path apps/majordomus-cli/Cargo.toml --test economics
• majordomus economics check

Požadované důkazy

• test_run_records_answer_the_task_shape

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/I3382.yaml. Read it back with majordomus plan show I3382.