What problem does it solve? It prevents the common debugging failure of patching symptoms with guesswork fixes by enforcing a strict rule: no fix without a verified root cause, so bugs stop recurring after repeated failed patches. ## Core Features & Use Cases - Root-cause-first workflow: An eight-phase process covering symptom collection, reverse code tracing, recent-change review via git, reproduction, testable hypothesis, verification, minimal fix, and structured reporting. - Fix authorization gate: Distinguishes diagnose-only mode (read-only investigation, default for ambiguous reports) from fix-authorized mode (explicit fix intent or verified scoped continuation), so it never modifies files without permission. - 3-strike rule: After three failed hypotheses or unverified patch attempts, it stops stacking patches and re-examines assumptions or architecture instead. - Use Case: When a user reports "this test is flaky, debug it", the skill gathers the exact error, reproduces the failure, checks recent git history, verifies a concrete root cause hypothesis, and reports evidence before any code change. ## Quick Start Ask the AI to investigate why a specific bug or error occurs and verify the root cause before fixing it.
