What problem does it solve? When facing a bug, test failure, or performance issue, the temptation is to guess at a quick fix — which often treats symptoms instead of causes and leads to repeated failed attempts. This Skill enforces a disciplined root-cause-first debugging method so fixes actually stick. ## Core Features & Use Cases - Four-phase systematic method: Root cause investigation, pattern analysis, single-hypothesis testing, then verified implementation — with an Iron Law of no fixes before investigation. - Repository evidence integration: Reads the repo's code map landmines, schema notes, code graph, and verification map so Phase 1 starts from recorded failure modes instead of an empty context. - Supporting techniques: Includes root-cause tracing through call stacks, defense-in-depth validation layers, condition-based waiting to replace flaky timeouts, and a bash script that bisects a test suite to find which test pollutes shared state. - Use Case: A payment test fails intermittently despite added sleep delays. Instead of increasing the timeout again, the Skill walks you through reproducing reliably, tracing the bad value to its source, forming one hypothesis, and verifying a minimal fix with the repo's mapped check commands. ## Quick Start Ask the AI to debug a failing test or bug using the systematic debugging skill before proposing any fix.
