What problem does it solve? When a FastAPI backend returns 404, 500, or timeout errors after deployment, developers often guess at fixes and break unrelated code. This Skill enforces a disciplined hypothesis-driven diagnostic workflow that locates the true root cause before any modification, applies minimal diffs, and verifies every endpoint afterward. ## Core Features & Use Cases - Hypothesis-Driven Diagnosis: Generates ranked root-cause hypotheses (missing routes, middleware ordering, schema mismatch, stale processes, dependency injection failures) with explicit verification methods before touching code. - Intensity Routing (L1/L2/L3): Controls reasoning depth per step — mechanical checks run fast, root-cause analysis gets full reasoning, preventing wasted tokens and premature conclusions. - Minimal-Diff Repair with Regression: Applies alias-route delegation patterns, backs up files before editing, then re-tests all registered endpoints including long-inference timeout scenarios. - Use Case: A frontend call to /api/chat returns 404 after a redeploy. The Skill confirms the process is alive, discovers the route is registered as /api/chat/ask, adds an alias route delegating to the existing handler, and verifies all endpoints return 200. ## Quick Start Diagnose why my FastAPI endpoint /api/chat returns 404 and fix it with minimal changes, then verify all other endpoints still work.
