What problem does it solve? When code crashes, tests fail, endpoints slow down, or security gaps appear, teams often patch symptoms instead of causes. This Skill enforces an evidence-first diagnostic workflow that proves the root cause before any code is modified, and gates remediation behind explicit user approval. ## Core Features & Use Cases - Root Cause Analysis: Reproduce failures, isolate causes via stack traces and git history, and record hypotheses backed by linked evidence before proposing fixes. - Multi-Domain Audits: Run focused investigations with --focus quality|security|performance|architecture covering secrets detection, injection risks, N+1 queries, algorithmic complexity, and structural code smells. - Gated Remediation: Apply fixes only with --fix after the user approves a ranked remediation plan with blast-radius ratings, then verify by re-running tests. - Use Case: When an API endpoint feels slow, invoke the skill with a performance focus to profile hot paths, record the findings in an evidence ledger, and receive ranked fix options rated Low/Medium/High risk before any edit occurs. ## Quick Start Ask the AI to run /do-diagnose on the failing endpoint or test with the appropriate focus flag to get an evidence-backed root cause and ranked fix options.
