How to audit an AI agent skill: the 7-check framework we used on 200 skills (opens in new tab)

{/* JSON-LD generated server-side in app/blog/[slug]/page.tsx; inline {...} blocks crash MDX's Acorn parser on the leading {. */} TL;DR This is the full methodology we use to audit AI agent skills (Claude Code, Cursor, Codex CLI, Gemini Code Assist) before we ship them in the Read this if: You are a solo developer who installs skills from random GitHub repos and wants to do that more safely. You are a security engineer on a small team being asked "are these skills OK to use." You are building...