Phase F.1: skill-eval engine (MVP) + framework fixes#158

Merged

azalio merged 12 commits into

mainfrom

Jun 3, 2026

Commits on Jun 3, 2026

feat(skills_eval): add eval_schema data contracts (ST-002)

azalio
and
claude
committed
feat(skills_eval): add VariantDispatcher + Mock/ClaudeSubprocess dispatchers (ST-003)

azalio
and
claude
committed
feat(skills_eval): add deterministic assertion runner (ST-004)

azalio
and
claude
committed
feat(skills_eval): add prompts×runs matrix runner with durable resume (ST-005)

azalio
and
claude
committed
feat(skills_eval): add aggregator stats + bounded-concurrency runner (ST-006)

azalio
and
claude
committed
feat(cli): register `mapify skill-eval run` subcommand (ST-007)

azalio
and
claude
committed
feat(skills): ship map-skill-eval skill via single-source render (ST-008)

azalio
and
claude
committed
test(skills_eval): dispatcher/assertion/anti-anthropic/e2e tests + fixture (ST-009)

azalio
and
claude
committed
docs(map-skill-eval): correct eval-set format + --max-concurrency default

azalio
and
claude
committed
docs(map-efficient): note record_test_baseline returns compact JSON — no head/tail

azalio
and
claude
committed
fix(framework): two MAP self-bugs surfaced during the F.1 run

azalio
and
claude
committed
docs(learning): capture Phase F.1 patterns + framework-error rule

azalio
and
claude
committed