About us
Helping AI agents get real work done
LLM SkillHub is an open platform where developers find, share and measure skills for AI agents. A skill teaches an agent a task once, so it does not have to work it out again, and spend tokens again, every time.
Why we built it
AI agents are smart. They are not yet practised.
AI agents like Claude Code, Cursor and Codex can now write code, research, analyse data and run whole workflows. But every time they start a task, they start from scratch. They work out the same steps again, make the same mistakes again, and spend the same tokens again. People end up pasting the same instructions into chat after chat.
Skills fix that. A skill is a small package, a SKILL.md file with instructions plus optional scripts and reference files, that an agent loads when a task calls for it. Written once, it makes the agent faster, more reliable and cheaper to run, every time after.
But skills were scattered across GitHub repositories, hard to find, and installed on trust. An agent follows whatever its skills tell it, so installing one written by a stranger is a real risk.
We built LLM SkillHub to be the place where skills are easy to find, safe to install and simple to share.
What we do
One home for agent skills
Find skills by what you need
Describe the problem in plain words and get the skill for it, even when none of your words appear in its name.
Checked before it is listed
Every public version is unpacked, scanned for leaked passwords and hidden instructions, and read by a person before anyone can install it.
Install in one line
From the website, from the llmsh command line, or straight from your agent over MCP.
Evals to measure skills
Test cases that show how well a skill does its job, so you can choose between skills on evidence.
Private spaces
Keep skills to yourself or your team, and share them with exactly the people you choose.
Versions you can compare
Every release is versioned, and any two versions can be compared side by side before you upgrade.
How it works
From a folder to every agent
- 1
Write it once
Put what you know into a SKILL.md file: when the skill should be used, the steps, and the mistakes to avoid.
- 2
Publish it
Upload it on the website or run llmsh publish. Public skills are checked and reviewed before they are listed.
- 3
Use it everywhere
Anyone can add it to Claude Code, Cursor, Codex, Gemini CLI, opencode, or any agent that speaks MCP.
What we believe
The principles we build by
Open, not locked in
Skills use the open Agent Skills format. A skill published here works across agents, and stays yours.
A person looks first
Automated checks catch a lot. A person reading the instructions catches the rest. It is slower, and we think it is worth it.
Credit where it is due
Some skills are copied from public repositories. They keep their authors’ licences, say so on the page, and come down on request.
Free to use
Finding, installing and publishing skills on LLM SkillHub is free.
Who we are
Founded by 2 engineers who suffered from the same problems
LLM SkillHub was founded and is run by its two co-founders, who build the platform, operate it and review what goes into it.
Priyanshu Jha
Get in touch