Thanks for your interest in contributing! This project follows the "rules from mistakes" philosophy — every addition should come from real experience, not theory.
- New skills — Structured workflows for common development tasks
- Stack templates — CLAUDE.md configurations for popular tech stacks
- Hook examples — Scripts for Claude Code's hooks API (PreToolUse, PostToolUse, Notification, Stop, SubagentStop)
- Bug fixes and improvements to existing content
- Documentation clarity improvements
- MCP server examples
-
Create
skills/your-skill-name/SKILL.md -
Follow this structure:
- Name — Clear, descriptive
- When to Use — Specific triggers
- Core Principle — One sentence philosophy
- Process — Numbered steps (5-7 max)
- Output Format — What Claude should produce
- Rules — Hard constraints
- Anti-Patterns — What NOT to do
-
Every rule must come from real experience. Include the mistake that inspired it.
- Create
examples/claude-md-templates/your-stack.md - Include: project structure, key files, verification commands, stack-specific rules
- Test it with a real project first
- Create
examples/hooks/your-hook.sh - Specify the hook type in the filename or header comment
- Include error handling and logging
- Test on both macOS and Linux if possible
- No theoretical rules — Every rule must come from a real mistake or problem
- No bloat — If it doesn't add clear value, don't add it
- Test first — Try your contribution with a real Claude Code project before submitting
- Keep it simple — Prefer clarity over cleverness
- Fork the repo
- Create a feature branch (
git checkout -b add-rust-template) - Make your changes
- Test with a real Claude Code project
- Submit a PR with a clear description of what problem this solves
Be respectful, be helpful, be specific. We're all here to make Claude Code better.