huggingface/hf-mcp-server

What it solves

This project provides an official Model Context Protocol (MCP) server that allows Large Language Models (LLMs) to interact directly with the Hugging Face Hub and thousands of Gradio AI applications. It bridges the gap between an LLM's reasoning capabilities and the vast ecosystem of models, datasets, and tools hosted on Hugging Face.

How it works

The server implements the MCP specification, supporting both STDIO and Streamable HTTP (JSON RPC) transports. It exposes Hub API and Search endpoints as tools that an LLM can call. It can also proxy other MCP servers via a CSV configuration and integrate with Gradio AI applications. The server includes a management web interface for monitoring server status, tool-call statistics, and "Skills" activity.

Who it’s for

• LLM Users: Those using MCP-compatible clients like Claude Desktop, Claude Code, Gemini CLI, VSCode, or Cursor who want to give their AI assistants access to Hugging Face resources.
• Developers: Those looking to deploy their own MCP servers or integrate Hugging Face Hub functionality into AI agents.

Highlights

• Broad Integration: Official support for installation in major AI IDEs and assistants (Claude, Gemini, Cursor, VSCode).
• Flexible Transport: Supports both local STDIO and stateless Streamable HTTP JSON mode.
• Extensible: Ability to load proxy tools from external MCP endpoints via CSV.
• Hub Access: Provides tools for repository search, filesystem access, and dynamic Space interaction.
• Monitoring: Built-in management dashboard for tracking metrics and tool usage.