소개
This skill implements a strict 'evidence-before-assertions' protocol for Claude Code, preventing the AI from making premature success claims or submitting unverified work. By mandating a specific gate function—identifying the verification command, running it, and confirming the actual output—it eliminates common failures like 'it should work' assumptions or relying on outdated test results. It is particularly useful for maintaining high code quality, preventing regressions, and ensuring that all commits and pull requests meet the defined requirements with 100% certainty.