What problem does it solve? Debugging often devolves into guesswork where developers patch symptoms instead of root causes, apply multiple unverified changes at once, or refactor unrelated code. This Skill enforces a structured diagnostic-then-fix workflow that isolates the true root cause before any code is modified. ## Core Features & Use Cases - Structured Six-Phase Workflow: Gathers symptoms, traces root causes with the Ariya-4 problem frame (Problem, Cause, Cessation, Path), confirms hypotheses via reproduction, applies minimal fixes, and verifies results. - GitNexus Knowledge Graph Integration: Uses query, context, trace, and detect_changes to map call chains, inspect suspect symbols, and measure the blast radius of every fix. - Hallucination Guardrails: Built-in sanity checklists prevent common failure modes like fixing symptoms, editing in loops without verification, or bundling multiple fixes. - Use Case: When a payment endpoint returns intermittent 500 errors, the workflow traces the call chain from the handler to an external API call missing a timeout, applies a single targeted fix, and confirms only the expected file changed. ## Quick Start Invoke the diagnostic-fix workflow with an error message or stack trace, such as asking the agent to diagnose why the checkout endpoint intermittently returns 500 errors.
