How to use Claude skills: install one, trigger it and check it worked
How to use Claude skills in four steps: get a skill folder from a source you trust, read its files, place or install it where Claude looks for skills, then ask for a task that matches its description or call it by name. A skill is a folder with a SKILL.md file of instructions, plus optional scripts and resources, that Claude loads only when it is relevant.
Skills save you from pasting the same instructions into every chat. Once a skill is installed, Claude sees its short description all the time and reads the full instructions only when a request calls for them.
Key ideas
- A skill is a folder with a SKILL.md file: a name and description at the top, instructions below.
- In Claude Code, personal skills live in ~/.claude/skills/ and project skills in .claude/skills/; claude.ai takes uploaded zip files.
- Claude uses a skill when your request matches its description, or when you type a slash and the skill name.
- Install skills only from sources you trust, and read every file first, including scripts.
What a skill is
Anthropic describes skills as folders of instructions, scripts and resources that Claude loads dynamically for specialised tasks. Every skill has a SKILL.md file that opens with a short YAML block: a name and a description of what the skill does and when to use it. The instructions follow in plain Markdown.
Loading happens in stages. The name and description, about 100 tokens per skill, are loaded at startup. The body of SKILL.md enters the conversation only when a request matches. Extra files, such as reference notes or scripts, are read or run only when the instructions point to them, so a large skill costs almost nothing until it is used.
Where skills live and how to install one
In Claude Code a skill is a folder on disk. Put it in ~/.claude/skills/<skill-name>/ to use it in all your projects, or in .claude/skills/<skill-name>/ inside a repository to share it with everyone working there. Skills can also arrive inside plugins: Anthropic's own collection installs with /plugin marketplace add anthropics/skills, then /plugin install document-skills@anthropic-agent-skills.
In claude.ai, custom skills are uploaded as zip files under Settings, Features, on paid plans with code execution turned on. On the Claude API they are uploaded through the Skills endpoints and run with the code execution tool. Skills do not sync between these surfaces, so install them separately in each place you use them.
How to trigger a skill and check it worked
Claude matches each request against the descriptions of installed skills, so asking for the task in words close to the description is usually enough. To call a skill directly in Claude Code, type a slash and its name; plugin skills use the form /plugin-name:skill-name. Arguments can follow the name.
To check, type a slash to see the command menu or run /skills to list the skills loaded in the session. If a skill never fires on its own, its description is probably too vague: it should say both what the skill does and when to use it. Then test it on a real task, since Anthropic notes that its example skills are for demonstration and may behave differently in your environment.
Safety, and a catalogue to choose from
Anthropic's guidance is to use skills only from trusted sources: ones you wrote yourself or got from Anthropic. A skill can direct Claude to call tools and run code, so audit every bundled file for unexpected network calls or file access. In Claude Code, also review a skill's allowed-tools field and any shell commands it runs before Claude reads it; treat installing a skill like installing software.
VITON13 SCHOOL keeps a free catalogue of 71 skills from GitHub on 11 shelves. Each entry shows its GitHub star count, which measures popularity rather than quality, a short note on what the skill does and the command to install it.
In everyday language
A skill is like a recipe card in a kitchen drawer. The cook sees the titles all the time, pulls out the full card only when that dish is ordered, and you should read any card a stranger hands you before it goes in the drawer.
Try it yourself
Open the skills catalogue linked below, pick one skill from the shelf closest to your work and open its GitHub repository. Read its SKILL.md and any scripts, then install it with the listed command and run /skills in Claude Code.
Expected result
The new skill appears in the /skills list, and a request that matches its description makes Claude load it.
Check your answer: A skill is installed and listed, but Claude never uses it on its own. What is the first thing to check?
Its description in SKILL.md. Claude decides from the description alone, so it must say what the skill does and when to use it; a direct slash call also confirms the skill itself works.
Questions
Do I need to code to understand how to use Claude skills?
No for installing and triggering them: a skill is a folder and a Markdown file. Reading the bundled scripts before you install a skill is easier with some coding knowledge, and that review is the part not to skip.
Does a skill installed in Claude Code also work in claude.ai?
No. Claude Code skills are files on your computer, claude.ai skills are uploaded in settings and API skills are uploaded through the API. Each surface needs its own copy.
Choose one skill for a task you repeat, read its files, install it and confirm it with /skills. The free catalogue at VITON13 SCHOOL lists 71 skills by task with the install command for each.
Sources and further reading
- Extend Claude with skills, Claude Code docs ↗Sources checked:
- Agent Skills overview, Claude Platform docs ↗Sources checked:
- anthropics/skills repository, GitHub ↗Sources checked: