clauderig

Bootstrap a production-grade .claude/ setup into any project — one command.

pipx install clauderig
cd my-project
claude-setup init

What You Get

• settings.json — permissions, allowed commands, auto-lint hook, pre-configured MCP servers
• commands/ — slash commands: /claude-fit, stack-specific add/review commands
• skills/ — 2–3 skill folders with real guidance (not lorem ipsum)
• hooks/post-edit-lint.sh — auto-runs linter after every file edit
• hooks/setup-mcps.sh — one-shot MCP prerequisite installer
• rules/coding-standards.md — dos and don'ts for your stack

Supported Stacks

Install

# Recommended (isolated environment)
pipx install clauderig

# Or via pip
pip install clauderig

Requires Python 3.10+.

Usage

Initialize a project

# Interactive
claude-setup init

# Non-interactive
claude-setup init --lang python --framework fastapi --target .
claude-setup init --lang php --target .
claude-setup init --lang react --framework reactjs --force

List supported stacks

claude-setup list

Check version

claude-setup version

The /claude-fit Command

After claude-setup init, open your project in Claude Code and run /claude-fit.

Claude will:

1. Scan your dependency files (requirements.txt, package.json, composer.json, etc.)
2. Detect libraries you're actually using (SQLAlchemy, Redis, Stripe, etc.)
3. Propose new skills and commands tailored to your specific project
4. Update CLAUDE.md with discovered project context (test command, folder structure, auth approach)

This turns a generic .claude/ setup into one that knows your exact project.

MCP Servers

Each stack ships with MCP servers pre-configured in settings.json. Claude Code reads these automatically.

Run .claude/hooks/setup-mcps.sh once after init to install npm prerequisites.

Set GITHUB_TOKEN env var for GitHub MCP authentication.

Publishing to PyPI

# Build
python -m build

# Upload (test first)
twine upload --repository testpypi dist/*

# Upload to production
twine upload dist/*

Requires a PyPI account and twine (pip install twine).

License

MIT