I Built a Claude Code Skill That Refuses to Report Vulnerabilities It Can't Prove

TL;DR: I previously built a Claude Code skill that runs a full-stack security audit in one command — a pattern library that tells the AI where to look. Running