runplaybook

MDM for AI agents. Scaffold agent docs, auto-install skills matched to your tech stack, and provision org-wide agent profiles — all from one command.

npx runplaybook init

runplaybook.sh

What it does

1. Scaffolds agent docs into your project:

project/
├── AGENTS.md              # Agent instructions (read by Cursor, Copilot, Warp, 20+ tools)
├── CLAUDE.md              # Claude Code wrapper
└── docs/
    ├── AGENT_SETUP.md     # Setup guide for your selected agent type
    ├── ARCHITECTURE.md    # System diagram, directory structure, API reference
    └── CHANGELOG.md       # Customer-facing changelog

2. Detects your tech stack and auto-installs matching skills from the skills.sh registry into .agents/skills/. Supports Next.js, React, Vue, TypeScript, Tailwind, Supabase, Expo, Firebase, Rust, Go, Python, and more.

No stack detected? You’ll be prompted to scaffold your project first and re-run.

3. Adds a git init hook to ~/.zshrc so runplaybook init runs automatically on every new repo.

Commands

npx runplaybook init      # Scaffold docs + install skills
npx runplaybook check     # Verify agent docs are complete

Agent types

Choose your agent type on first run:

• Developer — For coding projects. Scaffolds AGENTS.md with tech stack, commands, architecture, and conventions. Includes git repo detection.
• Executive — For business users. Guides MCP server setup, tool connections, and agent permissions.

Tool compatibility

| File | Read by | |---|---| | AGENTS.md | Cursor, GitHub Copilot, Codex CLI, Windsurf, Devin, Amp, Warp, 20+ tools | | CLAUDE.md | Claude Code (imports AGENTS.md) |

License

MIT