Skills MCP: One Server, 7,000 AI Agent Skills

A new open-source project called Skills MCP aims to solve a growing pain for developers working with AI coding assistants: the fragmentation of skills and plugins across different tools. The project, which launched today on Hacker News, offers a single MCP server that aggregates roughly 7,000 AI agent skills, letting users search, preview, and install them directly into Cursor, Claude Code, Claude Desktop, Cline, and Windsurf.

The setup is straightforward—a single npx command:

npx -y @gengirish/skills-mcp

Once running, the server acts as a centralized registry. Instead of hunting through GitHub repos, documentation, or community forums for the right skill or plugin, developers can query the server to find what they need and then install it with minimal friction. The project is MIT-licensed and available on GitHub.

The timing is notable. As AI coding tools multiply and each adopts its own plugin or skill format, developers face an increasingly fragmented ecosystem. A skill built for Claude Code might not work in Cursor without modification, and vice versa. Skills MCP doesn't unify those formats—it simply makes them discoverable in one place—but for many developers, discovery is half the battle.

Early reaction on Hacker News has been positive. One commenter called it "amazing" and said it "solved my biggest problem," echoing a common frustration: knowing that a skill exists, but not being able to find it easily.

For now, Skills MCP is a young project with a single maintainer. Its success will depend on whether it can maintain a comprehensive, up-to-date index and whether the broader community embraces it as a standard discovery layer. But as an open-source bridge across an increasingly fragmented tool landscape, it's a promising step in the right direction.