The Agent Skills Directory
Open skills for AI agentsOne command to add agent skills
Highlights
What you can do with it
- A front-end engineer installs Vercel React best practices into Cursor
- A team adds React Native UI and CI/CD skills for an agent
- A security auditor installs code scanning and variant analysis skills
How to use it
- 1 Open skills.sh and search for the skill you need
- 2 Check that it supports the agent you use
- 3 Run the install command such as npx skills add owner/repo in your terminal
- 4 Activate or call the skill inside your agent
Who it is for
Pros
- Free hosting and distribution of skills
- One command installs, no manual prompt pasting
- Works across several major agents
Limitations
- Early-stage ecosystem, so coverage and quality vary by skill
- Skill packages are third-party code and instructions, so review before running
- Compatibility depends on the agent you use, and not every skill supports all of them
What The Agent Skills Directory is
The Agent Skills Directory, at skills.sh, is an open-source ecosystem of reusable capability modules for AI agents. The idea is that a skill carries procedural knowledge, meaning logic and a process, rather than a paragraph of prompt text you paste in each time. Installation is a single terminal command of the form npx skills add owner/repo, and the skills are described as compatible with mainstream agents including Gemini, Claude Code, GitHub Copilot, and Cursor. The directory ranks popular skills and publishes a 24-hour trend list, and developers can submit their own skill repositories. Hosting and distribution are free, and the project describes itself as being at an early stage of ecosystem building.
What you can do with it
- Install a skill into your agent with one terminal command
- Give an agent domain knowledge such as a best-practices set or a workflow
- Browse rankings and a 24-hour trend list to find popular skills
- Check whether a skill supports the agent you use
- Publish your own skill repository to the directory
Who it is for
The examples are front-end engineers installing Vercel React best practices so an agent generates code aligned with current standards, development teams adding React Native UI and CI/CD workflow skills, and security auditors installing code scanning and variant analysis skills. That points at developers who use AI coding agents daily and want consistent behavior, and at teams that want one agreed way for an agent to handle a repeated task. Anyone who has built a good prompt library and wants to package it is the third group.
What to watch out for
An early-stage ecosystem means uneven coverage and uneven quality. Some skills will be excellent and some will be thin or abandoned, so read the repository before installing. Skill packages are third-party code and instructions, which means you are executing something you did not write, so review it the way you would review a dependency. Compatibility is per-skill: not every skill supports every agent, and the listing names four agents rather than all of them. Finally, a skill improves how an agent works; it does not remove the need to check the output.
FAQ
How do I install a skill?
Run a command like npx skills add owner/repo in your terminal.
Which agents are supported?
Gemini, Claude Code, GitHub Copilot, and Cursor are named in the listing.
Is it free?
Yes, the platform currently provides free skill hosting and distribution.
What makes a skill different from a prompt?
Skills are packages containing logic and processes, described as procedural knowledge.
Why it is free
What the official page says： 免费
Free limits, trial length and features are decided by the official site and can change at any time.
Product preview
Screenshot archived from the listing page, for preview only.
