DEV Community

curatedmcp for CuratedMCP

Posted on • Originally published at curatedmcp.com

Hugging Face MCP: Search, Run, and Deploy 500k+ AI Models Directly from Your Agent

Install guide and config at curatedmcp.com

Hugging Face MCP: Search, Run, and Deploy 500k+ AI Models Directly from Your Agent

The Hugging Face MCP server connects your AI agent directly to the world's largest open-source AI model repository. Instead of switching tabs between Claude and the Hugging Face Hub, you can now search models, run inference, access datasets, and deploy Spaces entirely through conversation — making it dramatically faster to prototype with cutting-edge open-source models.

What It Does

This server unlocks full programmatic access to Hugging Face's ecosystem. Your AI agent can:

  • Search intelligently across 500,000+ models filtered by task (NLP, vision, audio), framework (PyTorch, TensorFlow), and performance metrics
  • Run inference on any model via the Inference API without local hardware
  • Access 150,000+ datasets with schema inspection and preview capabilities
  • Manage Spaces — deploy, update, and share interactive Gradio/Streamlit applications
  • Compare performance using the Open LLM Leaderboard and community benchmarks
  • Download weights and configs for fine-tuning or local deployment

If you're building with open-source models, this eliminates friction. No more copy-pasting model IDs or manually checking dataset schemas. Your agent becomes a Hugging Face power user.

How to Install

npx -y @huggingface/mcp
Enter fullscreen mode Exit fullscreen mode

Add to your Claude Desktop configuration:

{
  "mcpServers": {
    "hugging-face-mcp": {
      "command": "npx -y @huggingface/mcp"
    }
  }
}
Enter fullscreen mode Exit fullscreen mode

See the official documentation for Cursor, Windsurf, and advanced setup options.

Real-World Use Cases

  • Model Search & Evaluation: "Find me the best open-source sentiment analysis model under 1B parameters with recent updates. Show me the benchmarks and community discussions."
  • Inference on Demand: "Run this customer feedback through Mistral-7B and give me the top 3 themes." No local GPU required — inference happens via API.
  • Dataset Discovery: "What datasets exist for training a code completion model? Show me the schema and a sample of rows."

Full install guides for Claude Desktop, Cursor, Windsurf, and more at CuratedMCP.

Top comments (0)