What problem does it solve? A regression guard that has never been observed failing is only a claim, not evidence. This Skill proves a guard actually catches the defects it claims to catch by deliberately reintroducing each defect one at a time, running the full test suite, and confirming the failure output names that specific defect before restoring the fix. ## Core Features & Use Cases - Per-Defect Reintroduction: Reverts exactly one fix at a time so each failure can be attributed to a single named defect, never a combined ambiguous breakage. - Observed Failure Evidence: Requires an actual captured suite run whose output names the reintroduced defect, rejecting predictions or unrelated failures as proof. - Restore-and-Repass Cycle: Restores each fix immediately after confirmation and re-runs the suite to a passing state before moving to the next defect. - Use Case: After fixing three token-capture bugs and adding regression tests, use this Skill to revert each fix individually, capture the failing output naming each bug, restore, and record durable per-defect evidence that the guard bites. ## Quick Start Prove my regression guard works by reintroducing each defect it claims to catch one at a time and confirming the suite fails naming that defect before restoring the fix.
