Skip to content
Get packs

Guide · Agent skill

How to use the Claude Skills Meta-Skill agent skill

Updated · written with AI from the project's own documentation and checked against our catalogue data

The claude-skills meta-skill is an agent skill that teaches an AI agent how to build other agent skills. It is written for developers and teams who want to package a library, API, or body of documentation into a reusable skill folder. After reading this guide you will know what the skill does, how to install it, how to trigger it with plain requests, and what to check before you rely on the result.

Claude Skills Meta-Skill2025Emma/vibe-coding-cn · MIT

Quick start

  1. 1Download the ZIP. Download the claude-skills ZIP from the item page on this directory.
  2. 2Unzip into a skills folder. Unzip it into ~/.claude/skills/ for personal use or into .claude/skills/ inside a project.
  3. 3Or upload in the Claude apps. Open Settings → Capabilities → Skills and upload the ZIP there if you use the Claude apps.
  4. 4Describe your source material. Ask for a new skill and provide the docs, API spec, or code it should be built from.
  5. 5Check the result. Review the generated SKILL.md, confirm the name matches the folder, and run the quality checklist before using the skill.

What is an agent skill and what does this one do?

An agent skill is a folder of instructions that an AI agent can load when a task matches it. The entry file is SKILL.md, which begins with YAML frontmatter that gives the skill a name and a description.

When this meta-skill is installed, the agent follows its guidance to take scattered documentation, API specs, or code and package them into a skill folder. It also covers how to rewrite an existing skill that is too long, vague, or inconsistent, and includes a pre-delivery quality checklist. The skill itself does not perform a domain task such as answering questions about a particular library.

  • Turns docs, specs, repos, or code into a skill folder with a SKILL.md entry file
  • Adds optional reference notes, scripts, or templates where they help
  • Rewrites bloated skills so they trigger reliably
  • Runs a quality checklist before delivery
  • The skill file is written in Hindi

How do you install the claude-skills skill?

You can install the skill in one of three places. For a personal install, download the ZIP from the item page and unzip it into ~/.claude/skills/. For a single project, unzip it into the project's .claude/skills/ folder. In the Claude apps, you can instead upload the ZIP under Settings → Capabilities → Skills.

Once installed, the folder should contain SKILL.md. The item lists SKILL.md as the only file in the package.

  • Personal: unzip into ~/.claude/skills/
  • Project: unzip into .claude/skills/ inside your project
  • Claude apps: Settings → Capabilities → Skills, then upload the ZIP

How do you trigger the skill with a natural request?

You do not need a special command. Describe what you want in plain language and mention the kind of task. The skill's description covers requests to build a new skill from docs, specs, or a repo, to fix an existing skill that misfires, to split long content into references, and to add a validator or quality gate.

Be specific about the source material. Naming the docs, API spec, or code you have gives the agent what it needs to follow the extraction steps.

  • Create a new skill from this API documentation: [paste or link the material]
  • My existing SKILL.md is too long and triggers at the wrong times. Help me rewrite it.
  • Turn this library's README and three code samples into an agent skill with examples.
  • Split this long reference into a references/ folder with an index.md.

What does a finished skill folder look like?

The guide recommends a folder named after the skill, with SKILL.md at the root. Long-form material goes in references/, starting with an index.md for navigation. Scripts and templates are optional and go in scripts/ and assets/. The minimum valid version is SKILL.md alone.

The frontmatter name must use lowercase letters, digits, and hyphens, starting with a letter, and must match the folder name. The description should say what the skill does and when to use it, with concrete trigger keywords.

  • SKILL.md: required entry point with YAML frontmatter
  • references/index.md: recommended navigation index
  • scripts/: optional helpers
  • assets/: optional templates or static files
skill-name/ |-- SKILL.md |-- references/ | `-- index.md |-- scripts/ `-- assets/

What does the skill need and what are the safety notes?

The item states that the skill needs a Bash shell to run the validate-skill.sh and create-skill.sh scripts. Those scripts are mentioned in SKILL.md but are not included in the file provided with this item, so you will need to supply them yourself if you want to use the validation and generator commands. The skill does not run scripts itself and does not use the network, according to the item details.

The guide also tells the agent not to invent external facts. It does not check sources on its own, so you must verify any claims in the skills it helps you write.

  • Requires a Bash shell for the referenced scripts
  • Runs scripts: no
  • Uses network: no
  • Does not verify facts; check external claims against the original sources
./skills/claude-skills/scripts/validate-skill.sh skills/<skill-name> --strict

What people use it for

  • Packaging an internal API's documentation into a skill that other agents can load.
  • Rewriting a long, rarely triggered skill into a short SKILL.md with clear trigger conditions.
  • Splitting a large reference document into topic files with a navigation index.
  • Adding a pre-delivery checklist to a team's collection of skills for consistent structure.

Frequently asked questions

Is the claude-skills skill free to use?

The item is listed under the MIT licence. The item details do not mention any price, so check the licence text on the item page for the exact terms before reuse.

Do I need a Bash shell to use it?

The item says a Bash shell is required to run the validate-skill.sh and create-skill.sh scripts. Those scripts are referenced but not included in the file provided, so the validation and generator commands will not run until you add them.

How do I install claude-skills in Claude?

Download the ZIP, then either unzip it into ~/.claude/skills/ for personal use, into .claude/skills/ in a project, or upload it under Settings → Capabilities → Skills in the Claude apps.

Does this skill create facts or verify sources?

No. The guide tells the agent not to invent external details, but it does not check sources on its own. Verify any claims in the generated skill against the original documentation.

Why won't my skill trigger?

Check that the frontmatter name matches the folder name and uses lowercase letters, digits, and hyphens. Then make sure the description states what the skill does and when to use it, with concrete trigger keywords.

How is this different from an existing domain skill?

A domain skill performs a task in a specific area. This meta-skill only helps you create or repair skills, so it is the right choice when you are authoring rather than using a skill.

New agent skills, repos and AI tools — every week

The best new skills and open-source projects, the fastest-growing repos, and one free prompt. Five minutes to read. Unsubscribe anytime.

Our verdict

This skill is a good fit for developers who want to package a library or API into a reusable agent skill, and for teams that need consistent structure and quality checks across many skills. It is not useful if you only want to use an existing domain skill, or if you expect it to verify facts, since it relies on you to check sources.

Download the skill