A Free Claude Skill Builder and SKILL.md Checker

A skill is a folder with a SKILL.md file that teaches Claude a task, such as writing commit messages or following your API rules. Claude reads each skill’s name and description, and loads the full instructions only when a task matches. This free Claude skill builder writes the SKILL.md for you from a few fields, and its checker tests any skill against the rules in Anthropic’s documentation.

How to Build a Skill

1. Pick a quick start, or type a name, what the skill does and when to use it.
2. Choose where it is used: Claude Code, or an upload to claude.ai or the Skills API. Uploads accept fewer fields, so the builder hides the ones they would reject.
3. In Claude Code, choose where it is saved and who can start it: you and Claude, only you, or only Claude.
4. Write the steps and rules, and add an example and edge cases if they help.
5. Tick the tools Claude may use without asking, and open More options for arguments, a subagent, path limits and extra files.
6. Fix anything the skill check flags, then download the SKILL.md or the whole folder as a .zip.

How to Check a Skill

Switch to Check a skill, paste a SKILL.md and pick where it will be used. The report lists errors, warnings and tips with line numbers. Add the folder name to check that it matches the skill’s name.

What the Checker Looks For

• Frontmatter that starts on the first line, closes with —, and parses as YAML.
• A name of up to 64 characters, made of lowercase letters, numbers and single hyphens, that does not start or end with a hyphen.
• A description of up to 1,024 characters, written in third person, that says when to use the skill and has no XML tags.
• Misspelled fields, such as allowed_tools, that Claude Code ignores without an error.
• Fields that a claude.ai upload or the Skills API rejects. Only name, description, license, compatibility, metadata and allowed-tools are allowed there.
• A body under 500 lines, file links with forward slashes, and dates that will go out of date.
• Risky setups: a deploy or publish skill that Claude can start on its own, or allowed-tools that grant every shell command.

For every field and what it does, read our guide to the SKILL.md format. If a skill does not load when it should, see why a Claude Code skill is not triggering.

Where to Save a Skill

• .claude/skills/NAME/SKILL.md: this project only. Commit it so your team gets it.
• ~/.claude/skills/NAME/SKILL.md: all your projects on this computer.
• PLUGIN/skills/NAME/SKILL.md: inside a plugin, where it runs as /plugin:name.

Claude Code picks up new and edited skills without a restart. If the skills folder did not exist when the session started, run /reload-skills. Test a skill by asking something that matches its description, or by typing its name after a slash.

Built With Claude Code

We built this tool with Claude Code and took every rule from Anthropic’s Claude Code skills documentation, the Agent Skills specification and Anthropic’s skill writing best practices. No AI runs on this page. The builder fills in ready-written structure, and the checker runs plain rules in your browser, so it is free and works without an account.

Claude Skill Builder FAQ

Is the Claude skill builder free?

Yes. It is free, needs no sign-up and has no limits.

Does it send my skill anywhere?

No. Everything stays in your browser. Your answers are saved in this browser only, so they are still there next time.

What is the difference between a skill and CLAUDE.md?

CLAUDE.md loads in every session, so it suits short facts about your project. A skill loads only when it is used, so it suits procedures and long reference material. Our CLAUDE.md Generator builds the other file.

Why does my skill not run on its own?

Claude decides from the description. Say what the skill does and the words a user would type, such as “Use when the user asks for a commit message”. A skill with disable-model-invocation: true runs only when you type its name.

Can I use the same skill in Claude Code and on claude.ai?

Yes, if its frontmatter uses only the six fields an upload accepts. Claude Code reads those six too. Live command lines and $ARGUMENTS work only in Claude Code.