Claude Skills MCP

🚀 Claude Skills MCP Server

Leverage Claude's powerful new Skills system with ANY AI model or coding assistant, including Cursor, Codex, GPT-5, Gemini, etc. This MCP server extends Anthropic's Agent Skills framework across the entire AI ecosystem via the Model Context Protocol.

A Model Context Protocol (MCP) server that offers intelligent search capabilities for finding relevant Claude Agent Skills using vector embeddings and semantic similarity. This server implements the same progressive disclosure architecture described in Anthropic's Agent Skills engineering blog, making specialized skills accessible to any MCP-compatible AI application.

An open-source project by K-Dense AI - the creators of autonomous AI scientists for scientific research.

This MCP server enables any MCP-compatible AI assistant to intelligently search and retrieve skills from our curated Claude Scientific Skills repository and other skill sources like the Official Claude Skills.

🚀 Quick Start

For Cursor Users

Add through the Cursor Directory, or add to your Cursor config (~/.cursor/mcp.json):

{
  "mcpServers": {
    "claude-skills": {
      "command": "uvx",
      "args": ["claude-skills-mcp"]
    }
  }
}

The frontend starts instantly and displays tools, automatically downloading and starting the backend in the background (~60 - 120s due to RAG dependencies, one-time). Subsequent uses are instant.

Using uvx (Standalone)

Run the server with default configuration:

uvx claude-skills-mcp

This starts the lightweight frontend which auto-downloads the backend and loads ~90 skills from Anthropic's official skills repository and K-Dense AI's scientific skills collection.

With Custom Configuration

# 1. Print the default configuration
uvx claude-skills-mcp --example-config > config.json

# 2. Edit config.json to your needs

# 3. Run with your custom configuration
uvx claude-skills-mcp --config config.json

✨ Features

• Two-Package Architecture: The lightweight frontend (~15 MB) starts instantly, while the backend (~250 MB) downloads in the background.
• No Cursor Timeout: The frontend responds in <5 seconds, solving the timeout issue.
• Semantic Search: Utilizes vector embeddings for intelligent skill discovery.
• Progressive Disclosure: Implements multi-level skill loading (metadata → full content → files).
• Zero Configuration: Works out of the box with curated skills.
• Multi-Source: Loads skills from GitHub repositories and local directories.
• Fast & Local: No API keys needed, with automatic GitHub caching.
• Configurable: Allows customization of sources, models, and content limits.

💻 Demo

Semantic search and progressive loading of Claude Agent Skills in Cursor

📚 Documentation

• Getting Started - Covers installation, Cursor setup, CLI usage, and troubleshooting.
• Architecture Guide - Details the two-package design, data flow, and components.
• API Documentation - Provides tool parameters, examples, and best practices.
• Usage Examples - Includes advanced configuration, real-world use cases, and custom skill creation.
• Testing Guide - Offers complete testing instructions, CI/CD, and coverage analysis.

🛠️ MCP Tools

The server provides three tools for working with Claude Agent Skills:

1. find_helpful_skills - Conducts semantic search for relevant skills based on task description.
2. read_skill_document - Retrieves specific files (scripts, data, references) from skills.
3. list_skills - Views the complete inventory of all loaded skills (for exploration/debugging).

See API Documentation for detailed parameters, examples, and best practices.

🔧 Technical Details (v1.0.0)

The system uses a two-package architecture for optimal performance:

• Frontend (claude-skills-mcp): A lightweight proxy (~15 MB).

  • Starts instantly (<5 seconds) ✅ No Cursor timeout!
  • Auto-downloads the backend on first use.
  • Serves as an MCP server (stdio) for Cursor.

• Backend (claude-skills-mcp-backend): A heavy server (~250 MB).

  • Performs vector search with PyTorch & sentence-transformers.
  • Functions as an MCP server (streamable HTTP).
  • Can be auto-installed by the frontend or deployed standalone.

Benefits:

• ✅ Solves the Cursor timeout issue (frontend starts instantly).
• ✅ Offers the same simple user experience (uvx claude-skills-mcp).
• ✅ Allows the backend to download in the background (doesn't block Cursor).
• ✅ Enables connection to a remote hosted backend (no local install needed).

See Architecture Guide for detailed design and data flow.

📦 Skill Sources

Load skills from GitHub repositories (direct skills or Claude Code plugins) or local directories. By default, it loads from:

• Official Anthropic Skills - 15 diverse skills for documents, presentations, web artifacts, etc.
• K-Dense AI Scientific Skills - 78+ specialized skills for bioinformatics, cheminformatics, and scientific analysis.
• Local directory ~/.claude/skills (if it exists).

🤝 Contributing

Contributions are welcome! To contribute:

1. Report issues: Open an issue for bugs or feature requests.
2. Submit PRs: Fork, create a feature branch, ensure tests pass (uv run pytest tests/), then submit.
3. Code style: Run uvx ruff check src/ before committing.
4. Add tests: New features should include tests.

Development

Version Management: This monorepo uses a centralized version system:

• Edit the VERSION file at the repo root to bump the version.
• Run python3 scripts/sync-version.py to sync all references (or use --check to verify).
• The scripts/build-all.sh script automatically syncs versions before building.

For questions, email orion.li@k-dense.ai

🌟 Join Our Community!

We'd love to have you in our Slack community! Connect with other users, share tips and tricks, get help with your skills, and be the first to know about new features and updates. 👉 Join the K-Dense Community on Slack 👈

Whether you're building custom skills, integrating with different AI models, or just exploring the possibilities of Agent Skills, our community is here to support you!

📖 Learn More

• Agent Skills Documentation - Official Anthropic documentation on the Skills format.
• Agent Skills Blog Post - Announcement and overview.
• Model Context Protocol - The protocol that makes cross-platform Skills possible.
• Engineering Blog: Equipping Agents for the Real World - Technical deep-dive on the Skills architecture.

📄 License

This project is licensed under the Apache License 2.0. Copyright 2025 K-Dense AI (https://k-dense.ai)

⭐ Star History