This skill establishes a rigorous 'Iron Law' for Claude Code, preventing premature or false completion claims by requiring fresh, documented verification evidence before any success assertions are made. It implements a specific gate function that forces the identification and execution of verification commands, analysis of output, and confirmation of results, ensuring that builds, tests, and bug fixes are truly validated before PR creation or handoff. By eliminating speculative language and relying on hard evidence, it builds trust and prevents the shipment of broken or incomplete features.
主な機能
01Prevention of speculative language like 'should' or 'probably'
02Validation of agent-delegated tasks through VCS diff checks
03Mandatory 'Iron Law' verification gate for all claims
04Strict red-green TDD cycle enforcement for regression tests
05Evidence-based success reporting requiring command output
060 GitHub stars
