A Claude Code skill is a folder with a SKILL.md of instructions that Claude loads when your request matches its description, or when you type /skill-name. Commit it to .claude/skills/ and everyone who clones the repo gets it.
A Claude Code skill is a SKILL.md file of instructions that Claude adds to its toolkit, plus any scripts, templates or reference files you bundle with it. Claude reaches for a skill when your request matches it, or you run it yourself as /skill-name. Skills follow the open Agent Skills standard, which Codex skills use too, and custom slash commands have been merged into them: .claude/commands/deploy.md and .claude/skills/deploy/SKILL.md both create /deploy. The built-in commands a team leans on are in Claude Code commands.
Unlike CLAUDE.md, which is loaded into every session, a skill's body loads only when it is used. A team can keep many skills without paying for all of them in every prompt. And unlike a hook, which runs every time its event fires, a skill runs when Claude decides it fits.
Where Claude Code skills live
Each skill is a folder with a SKILL.md inside. Where the folder sits decides who gets it:
Personal: ~/.claude/skills/<skill-name>/SKILL.md, available in every project on your machine.
Project: .claude/skills/<skill-name>/SKILL.md, committed with the repository, so everyone who clones it gets the skill.
Plugin: <plugin>/skills/<skill-name>/SKILL.md, installed with the plugin and run as /plugin-name:skill-name.
Enterprise: in the managed settings directory, set by an administrator.
How Claude decides to use a skill
Skills load in three levels, which Anthropic calls progressive disclosure. That is why the description matters more than anything else in the file:
The name and description are always loaded (about 100 tokens per skill), and Claude matches your request against the description.
The SKILL.md body loads only when the skill is triggered; the docs suggest keeping it under 500 lines.
Bundled files are read only when needed, and a bundled script's code never enters the context. Only its output does.
How to write a skill your team will reuse
Anthropic's skill authoring guide comes down to a few rules that matter more on a team, where a skill has to work for people who did not write it:
Write the description for the match: what the skill does and when to use it, in the third person. It is the only part Claude sees before deciding.
Keep SKILL.md short and move detail into files it links to, one level deep.
Install skills only from sources you trust: a skill can run code with your permissions, and a malicious one can exfiltrate data.
See which skills your team actually uses
An installed skill is not a used one. In one engineer's 323 Claude Code sessions (September 2 to October 3, 2026), a skill loaded in 49 (15%). Thirteen skills loaded 58 times: two browser-automation skills made up 25 of them, a design-review skill and a charting skill came next, and the Stripe plugin, enabled the whole time, used none of its 10 skills or its MCP tools.
Zest's Claude Code plugin records every Skill tool call and every SKILL.md a session loads, from project, personal and plugin folders. The AI Stack view lists each engineer's skills next to their models, agents and MCP tools: which skills spread across the team, which stay on one laptop, and which sessions that used them ended in a merged pull request. The rest of what it records is on Claude Code usage.
TRY IT ON YOUR TEAM
See what your team ships with AI.
Starter is free (3 seats, no credit card). Pro is $49 a month with 5 seats and Max is $149 a month with 10 seats — priced per workspace, not per engineer, with extra seats at $19 a month.
The file that defines a Claude Code skill: optional YAML frontmatter (description is the one field the docs recommend), followed by Markdown instructions. It sits in a folder named after the skill, next to any scripts or reference files the skill uses.
How do I share a Claude Code skill with my team?+
Commit it to the repository under .claude/skills/, so everyone who clones the project gets it, or package it in a plugin and publish the plugin through a marketplace your team adds once.
Can I stop Claude from running a skill on its own?+
Yes. disable-model-invocation: true in the skill's frontmatter makes it manual-only: it runs when someone types /skill-name. user-invocable: false does the opposite and leaves it to Claude.
