Thanks for considering a contribution! This project is open to improvements of all kinds.
- New skills — add a
skills/<name>/SKILL.mdfollowing the existing format (YAML frontmatter + markdown body under 500 lines) - Improve existing skills — fix errors, add patterns, improve clarity
- Research additions — add benchmarks, comparisons, or new findings to
research/ - Bug reports — if a skill gives bad advice, open an issue
Every skill must have:
- A folder in
skills/with at least aSKILL.md - YAML frontmatter with
name(lowercase-hyphenated, max 64 chars) anddescription(max 1024 chars) - Body under 500 lines — use
references/for longer content - Agent/IDE-agnostic instructions (no vendor-specific commands)
- Be concise — every token costs context window space
- Explain why, not just what — LLMs respond better to reasoning than rigid rules
- Test with a real AI agent before submitting
- No secrets, API keys, or personal paths in any file
- Fork the repo
- Create a branch (
git checkout -b add-my-skill) - Make your changes
- Open a PR with a clear description of what the skill does and why
That's it. No CI to pass, no tests to run (yet). Ship it.