The Agent Skills Directory

[SAFE]: The skill implements a robust documentation workflow for Elixir/Phoenix development. It explicitly forbids editing system/plugin files and provides clear guidance on managing user-owned files like CLAUDE.md and package-specific skills. The provided references/common-mistakes.md is correctly treated as a read-only reference to prevent accidental overwrites. All file paths used (e.g., ~/.claude/skills/, .claude/skills/, and mix.lock) are within the expected scope for an AI agent assisting in development tasks. No suspicious network operations or command injections were identified.

learn-from-fix