This is an L2 coordinator that orchestrates quality evaluation for story-based workflows by delegating to four specialist workers: code quality checker, tech debt cleaner, regression checker, and test log analyzer. It enforces mandatory research (official docs, MCP Ref, Context7, web best practices) before any evaluation run, runs evidence gathering in parallel with agent review, then merges findings through a bounded refinement loop before rendering a PASS/CONCERNS/FAIL verdict. The phase sequencing is strict: read-only evidence first, cleanup second, agent barrier sync, then merge and refinement. It treats transport failures as operational noise rather than quality signals, which is a sensible design choice that prevents false negatives from rate limits or missing credentials. Built for teams running structured story pipelines with repeatable quality gates.
npx skills add https://github.com/levnikolaevich/claude-code-skills --skill ln-510-quality-coordinator