AGENTS.md instructions for ondrej-merkun/skill-audit, covering agents.md — entry point for any ai coding agent, read these in order, every session, spec & task references, build / test / lint commands (verbatim) and hard rules (loop-driver agnostic).
Claude Code instructions for ondrej-merkun/skill-audit, covering skill-audit, identity, tech stack, commands (exact strings — use verbatim) and directory map.
Ralph (and any human contributor) MUST read this file at the start of every session, before picking a task. When you finish a task and notice something that belongs here — a surprise, a near-miss, a "next time I'll…" — append it as a new numbered lesson in the appropriate section. One lesson per bullet, terse, with…
Scan installed agent skills, plugins, and cross-agent instruction files for prompt injection, exfiltration, malicious code, and dependency risk. Use when the user asks to audit, check, review, inventory, or verify skills for Claude Code, Codex, Cursor, Copilot, Gemini, or AGENTS.md-style project instructions.
Gives the agent read and write access to a set of allowed directories on the local filesystem. Runs locally from the @modelcontextprotocol/server-filesystem npm package.