Skills in Claude Code
Skills in Claude Code
Helps package repeatable workflows for development teams.
How to create, install and use your own skills for Claude Code, on your machine or per project.
Overview
This is the official guide to using skills in Claude Code, Anthropic's terminal-based AI coding tool. In Claude Code, a skill is a SKILL.md file containing instructions; Claude uses it automatically when it fits, or you call it directly with /skill-name.
Skills in Claude Code follow the open Agent Skills standard and add their own extensions such as controlling who can invoke a skill, running a skill in a subagent and injecting shell command output into the skill content. The old custom slash commands have been merged into skills; old files still work but skills are preferred for new work.
For development teams, skills are the most compact way to turn checklists or instructions you keep pasting over and over into reusable workflows that can be shared through git.
Highlights
Automatic or manual invocation
Claude loads a skill on its own when the request matches its description, or you type /skill-name with arguments.
Multiple storage scopes
Personal skills live in ~/.claude/skills/, project skills in .claude/skills/ and can be committed to share with the whole team.
Control who can invoke
disable-model-invocation: true lets only you call it; user-invocable: false hides it from the menu and leaves it for Claude only.
Inject dynamic data
The !`command` syntax runs a shell command and puts its output, such as git diff, into the skill content before Claude reads it.
Bundled skills
Claude Code ships with skills such as /code-review, /debug, /loop, /claude-api and /skill-doctor.
What it's for
Review code changes
Create a skill that summarizes git diff and points out risks for developers to check before committing.
Safe deploy process
Write a deploy skill that only a human can call manually, so AI doesn't push code to production on its own.
Standardize across the team
Commit the .claude/skills/ folder to the repo so every member shares the same way of writing tests, naming and writing commits.
Handle tickets by ID
Create a skill that takes arguments such as an issue number so Claude reads and handles it following your internal process.
How to use it
- 1
Create the skill folder
Create a folder for a personal skill in ~/.claude/skills/.
$mkdir -p ~/.claude/skills/summarize-changes - 2
Write SKILL.md
Create a SKILL.md file in that folder, with a frontmatter description stating clearly what the skill does and when to use it.
- 3
Add instructions
Write the steps Claude should take; you can use !`git diff HEAD` to insert current changes into the skill.
- 4
Run Claude Code
Open Claude Code in a git repo that has changes.
$claude - 5
Try calling the skill
Ask Claude what you just changed to try automatic invocation, or type the command to call it directly.
$/summarize-changes
Video guides
How to Use Skills in Claude Code (Create, Install & Manage Skills)
Code With Joel
How to Create Skills in Claude Code Claude SKILL.md (2026)
The Code City
Claude Skills: Full Giải Thích & Hướng Dẫn Tạo Skill A-Z
Phan Đông Giang - Học AI
Real-world example
Suppose a 5-person web development agency in Hanoi often forgets pre-handover checks. The team lead creates a project skill in .claude/skills/ describing a checklist: run tests, check environment variables, review the mobile interface. The deploy skill is set to disable-model-invocation: true so only someone who calls /deploy can run it. The skill folder is committed to the repo, so a new member who clones it can use the same process right away.
Tips & notes
Start with repeated work
When you notice yourself pasting the same instructions many times, turn them into a skill.
Lock sensitive skills
For deploys or high-impact actions, set disable-model-invocation: true so Claude doesn't call them itself.
Call it in the right place
Only a skill name at the start of a message runs directly; mentioning it mid-sentence only lets Claude use it.
Watch context cost
Use /skill-doctor to see how much context each skill takes and how it is used.
FAQ
Do skills replace the old slash commands?
Custom commands have been merged into skills. Files in .claude/commands/ still run, but use skills for new work; if names collide, the skill takes priority.
Do I need to restart after editing SKILL.md?
Changes inside skill folders are picked up within the session; if you create a new top-level skills folder after startup, run /reload-skills.
Are Word and PDF document skills built into Claude Code?
According to the Agent Skills docs, the prebuilt document skills are not available in Claude Code; you can install the versions from the anthropics/skills repository via a plugin.
How are skills in plugins invoked?
Skills that come from a plugin are invoked as /plugin-name:skill-name.
Useful links
Want to apply this in your business?
2dm looks at your workflows, picks the right tools and sets them up with your team. The first call is free.
Same topic
Anthropic Skills
4.5
Anthropic's official repository of example skills for Claude: each skill is a folder of instructions and resources that helps Claude do one specific kind of task well.
Agent Skills documentation
4.0
Explains what a skill is, the SKILL.md folder structure, and how Claude loads skills automatically when needed.
Introducing Agent Skills (article)
4.0
An article by Anthropic's engineering team on why Agent Skills exist and how they are designed.