01
What this skill does
Use this skill to resolve regression reports or related feedback. It verifies findings, graph evidence, intended changes, and coverage against current behavior and product requirements. It reconciles stale claims per item, continues authorized fixes, and preserves intended behavior and source feedback. Follow-up review is bounded to affected changes when justified. Invocation is explicit-only: a user must invoke `$receiving-regression-review`; matching prompts do not activate it automatically.
02
When to use it
- 01Re-checking a regression gate against the current diff and baseline.
- 02Reconciling behavior graph deltas with findings and coverage rows.
- 03Fixing only proven user-visible regressions.
- 04Separating real regressions from intentional product changes.
03
How it works
- 01
Read the report and requested outcome, preserving the source while listing findings, intentional changes, and uncovered surfaces.
- 02
Reconcile behavior graph deltas with the current code path and coverage ledger.
- 03
Verify current paths and product intent per item; reconcile stale claims without blocking independent fixes.
- 04
Create a disposition ledger before editing.
- 05
Fix confirmed regressions within existing authorization and run focused plus required verification.
- 06
Report every disposition and unresolved gate item; return justified follow-up findings without recursive receiving cycles.
04
What you get back
- 01A full disposition ledger.
- 02Scoped fixes for confirmed user-visible regressions.
- 03Evidence for challenged, narrowed, intentional, and carried-forward items.
05
Important boundaries
- 01Do not blindly apply review feedback.
- 02Preserve existing staged contents; perform Git actions only within the user's specific authorization for this task.
- 03Do not broaden user-visible behavior while resolving a focused regression finding.
06