Параллельный quality gate перед коммитом

★ 7.0 · testing

qc is a Claude Code skill that runs a quality gate with 5 parallel agents reviewing a git diff before a commit is allowed. Each agent covers a distinct dimension: Code Review (naming, bugs, debug statements), Simplification (over-engineering, YAGNI, dead code), Consistency (codebase patterns, type safety), Robustness (edge cases, regressions, API contracts), and Scope (whether changes stay on task). The first four agents use Claude Sonnet; the Scope agent uses Haiku. Every agent outputs raw JSON only — no markdown. All five checks must pass for the commit to proceed; any failure triggers a list of issues and prompts the developer to fix them and re-run `/qc`.