What problem does it solve? Bug fixes often ship without proof that the bug is actually resolved or that it will not regress later. This Skill enforces a test-driven bug-fixing workflow where the broken behavior is made executable as a failing test before any production code changes, giving concrete evidence the fix works. ## Core Features & Use Cases - Failing-Test-First Workflow: Guides a six-step process from understanding the bug through writing a failing regression test, fixing the code, and rerunning the test to confirm it passes. - Practical Fallbacks: When a real test is impractical (expensive harness, brittle mocks, slow end-to-end infrastructure), it substitutes the closest executable regression check such as a targeted script, manual reproduction command, or log assertion. - Guardrails Against Bad Tests: Prevents weakening existing assertions, mirroring implementation details in tests, or adding broad fixture churn unrelated to the bug. - Use Case: A user reports that canceling a run leaves orphaned child jobs. The Skill writes a focused unit test reproducing the orphan behavior, confirms it fails, applies the minimal fix, and confirms the test now passes. ## Quick Start Ask the AI to fix this bug using TDD by writing a failing regression test first, then making the minimal code change until the test passes.