What it is
MCP Skills Hub (formerly AI Library) is an open directory of 966 Model Context Protocol servers, Agent Skills and GitHub repositories for AI coding assistants. Every entry comes with the exact install configuration for Claude Code, Claude Desktop, Cursor, Codex, Windsurf, Cline and VS Code, so you can paste it rather than translate a README.
Who maintains it
It is built and maintained by Neel Bhavsar, and the code is open source on GitHub. It is independent and not affiliated with Anthropic, OpenAI, Cursor or any listed project.
Where the data comes from
A pipeline runs every morning. It reads the official MCP Registry, the official reference servers, Anthropic's and the community's skill marketplaces, and GitHub. It enriches each entry from npm, PyPI and the source repository: maintenance activity, download counts, release history, the tools a server exposes (asked of the live server, or read from its documentation), and a static scan of what a package's code can reach. Nothing is edited by hand except a short list of curated corrections kept in the repository.
How entries are judged
Each server gets a health score from published components and weights: maintenance, adoption, transparency, provenance and packaging. Anything that cannot be measured is left out rather than scored as zero. The score is not a security audit. The formula is on how the health score works, and it is open to argument.
Corrections
If a config does not work, a listing is wrong, or something should be added, the fastest route is a GitHub issue. The data is also available as a JSON API and an MCP server.