The open agent skills registry
$ ls skills/
Agent Skills — the open registry of installable skills for AI coding agents
19,296 installable · scanned daily
Reusable capabilities for your AI agents.
Install any of 19,296 skills with one command — works with every major coding agent. Free and open source.
Try it now
Available for these agents
How it works
From find to installed in under a minute.
- 01
Find a skill
Search the registry or browse by category, tool, or stack until you find the capability your agent is missing.
›browse · search · ⌘K - 02
Copy the command
Every skill page gives you a one-line install, tuned to your agent and whether you want it project-local or global.
›npx degit owner/repo … - 03
Drop it in
The skill installs as a SKILL.md into your agent's skills directory — .claude/skills, .codex/skills, and friends.
›→ .claude/skills/<skill> - 04
Your agent uses it
On the next run your agent picks the skill up automatically. No restart, no config, no registration step.
›# ready
Built on trust
Vet every skill before you install it.
An agent skill runs with your agent's permissions, so trust matters. We make the signals you need to decide impossible to miss.
Scanned for safety
Every skill gets a static safety scan with a clean / review / warn grade and capability flags — surfaced before you install, never hidden.
Source-linked & fresh
Each skill links to its GitHub source with stars, license, and last-commit date. We re-scan the whole catalog every day.
No lock-in
Skills are plain SKILL.md files. Fork them, edit them, vendor them into your repo — they're portable and they're yours.
About agent skills
Agent skills are reusable instruction sets that teach AI coding assistants new workflows, best practices, and domain knowledge. Drop one into your client's skills directory and your AI gains the relevant context immediately — no prompt engineering, no re-explaining.
Skills work with Claude Code, Codex, GitHub Copilot, Antigravity, and other AI assistants. Install with one command and start using immediately. Browse by category, check the leaderboard for the most popular skills, or pick a client-specific landing page.
What a skill actually contains
A skill is a folder with a SKILL.md file at its root: YAML frontmatter with a name and a description, followed by the instructions themselves. The description is what the agent matches against — it is the trigger that decides whether your request activates the skill. Supporting files (scripts, references, templates, assets) sit alongside it and are read only when the skill needs them.
How an assistant loads a skill
You install a skill into a client-specific directory — .claude/skills/<name> for Claude Code, for example, project or global scope. On each turn the agent sees the available skill names and descriptions, and loads the full instructions only when a request matches. That is why skills are cheaper than a long system prompt: the detail stays on disk until it is relevant.
How this registry is built
Every listing here is a real SKILL.md found in a public repository. Listings carry the verified source URL and the commit date the content was read from, and each skill page shows the actual file rather than a paraphrase, so you can judge the instructions before installing anything. Pages that cannot be traced to a live source are excluded from the index instead of being padded out. If a source moves, the listing is re-resolved or dropped.
Building something on top of the registry? The developer resources page documents the JSON API, the OpenAPI description, and how every page is also available as Markdown for agents.