Skip to content
Agent Skills OS

Contribute

Add a skill to the library.

Every skill here is one Markdown file someone wrote and shared. Fork the repo, add yours, and open a pull request. Fixes to existing skills and to the loader are just as welcome.

How to

Four steps, about ten minutes.

  1. Step 1

    Fork and clone

    Fork the repository on GitHub, then clone your fork and start a branch for your skill.

    terminal
    git clone https://github.com/<you>/agent-skills-os.git
    cd agent-skills-os
    git checkout -b add-my-skill
  2. Step 2

    Write the skill

    Create skills/<name>/SKILL.md. The folder name is the skill's name. Frontmatter first, then the instructions.

    skills/my-skill/SKILL.md
    ---
    name: my-skill
    description: When to use this skill and what it produces.
    version: 0.1.0
    tags: [domain, capability]
    inputs:
      - name: input_name
        description: What this input is.
        required: true
    related: [other-skill]   # optional
    author: your-name        # optional, only if you wrote it
    ---
    
    # my-skill
    
    The instructions the agent follows: the process,
    the output shape, and what to avoid.
  3. Step 3

    Check it loads

    Load it through the package and look at what comes back. Optionally write the Cursor, AGENTS.md and Stitch versions too.

    terminal
    node -e "import('./src/index.js').then(m => console.log(m.getSkill('my-skill')))"
    
    # Optional: Cursor, AGENTS.md and Stitch exports
    node scripts/export-skill.js my-skill
  4. Step 4

    Open a pull request

    Push your branch and open a pull request against main. Say what the skill is for and show one example of it working.

    terminal
    git add skills/my-skill
    git commit -m "Add my-skill"
    git push -u origin add-my-skill
    gh pr create --fill   # or open the PR on github.com

Guidelines

What makes a skill worth adding

How to write it

  1. Lead with intent. The first paragraph says when the skill applies.
  2. Process before output. Describe how to think, not only what to emit.
  3. Specify the output shape. Models drift without a target format.
  4. Name the failure modes. A "what to avoid" section prevents the common mistakes.
  5. Be concrete. One real example teaches more than a paragraph of advice.

What gets merged

  • One clear job per skill, named for what it does.
  • Your own writing. Add author and author_url only to a skill you wrote.
  • No secrets, keys or private data in the skill or its references.
  • Instructions a model can follow: a process, an output shape, and what to avoid.

Contributors

The people behind the library

Everyone who has landed a change in the repository. Your avatar shows up here after your first merged pull request.