What problem does it solve? When a bug, failing test, or build break appears, it is tempting to guess at fixes or rewrite large sections of code. This Skill enforces a disciplined workflow that finds the smallest causal explanation for an observed failure before any code is changed, preventing speculative fixes and unnecessary refactors. ## Core Features & Use Cases - Evidence-First Diagnosis: Records expected versus observed behavior and identifies the cheapest command that reproduces the difference. - Falsifiable Hypotheses: Forms one testable cause at a time and uses the cheapest discriminating evidence instead of broad audits or full test suites. - Minimal Causal Fixes: Fixes the causal seam, adds a focused regression test only when it protects the observed failure mode, and stops once verification passes. - Use Case: A CI build breaks after a merge. Instead of reverting blindly, use this Skill to reproduce the failure locally, isolate the exact input causing it, patch that seam, and re-run only the affected test. ## Quick Start Use systematic-debugging to find the root cause of this failing test and fix only the code responsible for it.
