Stop an AI agent damaging your repo: PreToolUse hooks, permission deny rules, protected paths, verification gates. Use when "claude keeps force pushing", "CLAUDE.md says X but it still does Y", "stop the agent touching prod or .env", or making a repo safe for unattended agent work. For AI features you ship to users use llm.
安装
npx skills add rainmanjam/poka-yoke --skill agent-guardrailsnpx skills add rainmanjam/poka-yoke --skill agent-guardrails -a claude-code -gnpx skills add rainmanjam/poka-yoke --skill agent-guardrails -a codex -ggemini skills install https://github.com/rainmanjam/poka-yoke.git --path plugins/poka-yoke/skills/agent-guardrailsgh skill install rainmanjam/poka-yoke agent-guardrailsgit clone --depth 1 https://github.com/rainmanjam/poka-yoke /tmp/rainmanjam__poka-yoke
mkdir -p ~/.claude/skills ~/.agents/skills
cp -R /tmp/rainmanjam__poka-yoke/plugins/poka-yoke/skills/agent-guardrails ~/.claude/skills/agent-guardrails # Claude Code
cp -R /tmp/rainmanjam__poka-yoke/plugins/poka-yoke/skills/agent-guardrails ~/.agents/skills/agent-guardrails # Codex, Cursor, Gemini CLI, Copilot, OpenCode