← ClaudeAtlas

fixing-codelisted

Fix code defects with a reproducible feedback loop, root-cause diagnosis, minimal patch, regression test, and clean verification. Use when debugging, diagnosing, or resolving lint/test/build failures. NOT for behavior-preserving refactors (use refactoring-code), test-suite cleanup without a production bug (use improving-tests), or code review findings without fixes (use reviewing-code).
alexei-led/cc-thingz · ★ 33 · Code & Development · score 81
Install: claude install-skill alexei-led/cc-thingz
# Fix and Diagnose Code Fix the requested defect or failing gate one verified issue at a time. Do not patch from guesses. Do not expand to unrelated failures without asking. Never use destructive git commands such as hard reset, clean, force push, or checkout-overwrites as a fix. ## Role-gated action Detect capability from tools: - Write-capable role: reproduce, diagnose, patch, test, and clean up. - Read-only role: diagnose from files and supplied output, then emit the fix in the Proposed Changes contract. Apply nothing; run nothing. - Missing key tool or permission: stop with Blocked and ask for the exact artifact, access, or approval needed. Use an interactive question tool when available for missing logs, payloads, repro steps, environment details, access, or permission for temporary instrumentation. ## Route elsewhere Do not use this for: - pure refactors with unchanged behavior → `refactoring-code` - test-only improvement or coverage work → `improving-tests` - review-only findings → `reviewing-code` - broad architecture redesign → architecture skills - browser-only UI investigation without a cheaper signal → `browser-automation` ## Reproduce first For lint/build/test failures, run the relevant project gate first. Prefer `make lint` and `make test` when present; otherwise use configured language tools from the nearest project root. For reported bugs, build the fastest reliable pass/fail signal: 1. Existing failing test or new regression test at the behavior