qa-slice

# QA Slice Replay affected baseline scenarios after an implementation wave and record the evidence in `.maestro/features/<id>/qa-slices.yaml`. The ship gate counts only slices with scenario ids and non-empty evidence. Activate: `maestro hook record --event skill_activation --skill qa-slice` ## Use - After a task wave changes feature behavior. - Before `maestro feature ship`. - When ship reports uncovered `[bl-NNN]` scenarios or stale QA evidence. ## Do 1. Read changed files/commands, `maestro feature show <id>`, and `.maestro/features/<id>/baseline.md`. 2. Select the affected `[bl-NNN]` scenarios. If the wave adds behavior, extend the baseline with a new id instead of hiding it behind a unit test. 3. Run the smallest useful probes: focused tests for local invariants, plus a real command/manual/API/UI flow when composition risk exists. 4. Compare against the baseline. Unexplained output, schema, state, permission, performance, compatibility, or UI drift is a blocker. 5. Append a counting slice to `qa-slices.yaml`. 6. If blocked, return a tracker entry with expected vs actual, reproduction, evidence, and fix path. Do not fix code from this skill. ## qa-slices.yaml Keep this append-only shape. Scenario ids must match baseline digits exactly: `bl-001` and `bl-1` are different. ```yaml slices: - at: "2026-05-31T00:00:00Z" scenarios: ["bl-001", "bl-002"] probes: ["cargo test --test feature_domain"] result: pass evidence: - "feature...