Skills for Claude Code. Each one is a folder
containing a SKILL.md — a set of instructions Claude loads when the task
matches, so you get the same behaviour every time instead of re-explaining
what you want.
Take any of them, change them, keep the parts you like.
| Skill | Invoke | What it does |
|---|---|---|
eli5 |
/eli5 <topic> |
Explains a topic from scratch in everyday words, using a fixed four-part shape, then hands you the real terminology so you can go deeper on your own terms. |
Every method below is a prompt you paste into Claude Code. It does the copying and tells you what landed. If you would rather run the commands yourself, each one has a shell equivalent.
Install the skills from https://github.com/nils-e/agent-skills into my
personal skills directory. Clone it to a temp location, copy each folder
under skills/ into ~/.claude/skills/, then tell me which skills are now
available and how to invoke them. Don't overwrite a skill I already have
with the same name — ask me first.
That last sentence matters. Without it, a skill you already have with the same name is replaced silently.
git clone https://github.com/nils-e/agent-skills.git
cp -r agent-skills/skills/* ~/.claude/skills/Name the folder you want:
Install just the eli5 skill from https://github.com/nils-e/agent-skills
into ~/.claude/skills/.
git clone https://github.com/nils-e/agent-skills.git
cp -r agent-skills/skills/eli5 ~/.claude/skills/A skill can be installed at two scopes, and the difference is who gets it.
| Scope | Directory | Available in | Good for |
|---|---|---|---|
| Personal | ~/.claude/skills/ |
Every project on your machine | Anything you want everywhere. The default. |
| Project | <repo>/.claude/skills/ |
That one repo | Team conventions — commit it and everyone who clones gets it |
Personal is the default above. For project scope:
Install the skills from https://github.com/nils-e/agent-skills into this
project's .claude/skills/ directory rather than my personal one, so I can
commit them for my team.
git clone https://github.com/nils-e/agent-skills.git
mkdir -p .claude/skills
cp -r agent-skills/skills/* .claude/skills/If the same skill name exists at both scopes, the project one wins inside that repo.
Start a new session, or restart Claude Code, and the skill is picked up. Type
/ to see it in the list.
A skill is a folder with a SKILL.md inside:
~/.claude/skills/
└── my-skill/
└── SKILL.md
SKILL.md is frontmatter plus instructions in plain Markdown:
---
name: my-skill
description: What this does and when to use it. Claude reads this to decide
whether the skill is relevant, so be specific about the trigger.
---
# My skill
The instructions. Write them as directions to follow, not as a description
of a feature.Two things matter more than the rest:
The description is the trigger. It is what Claude matches against the
task at hand, so say plainly when to use the skill, not just what it is. A
vague description means the skill never fires, or fires when you didn't want
it.
Write rules, not prose. "Lead with the answer, no preamble" is a rule. "This skill produces high-quality summaries" is a wish. Where a rule exists because something went wrong once, say what went wrong — it stops the rule being quietly dropped later.
A skill can also hold supporting files — reference docs, scripts, templates —
in its folder alongside SKILL.md, and point at them from the instructions.
Issues and pull requests are welcome. For a new skill: one folder under
skills/, a SKILL.md with real frontmatter, a README.md explaining what
it is for, and a row in the table above.
MIT. See LICENSE.