Hugging Face Spaces

MCP-HFSpace provides a bridge between Claude Desktop and Hugging Face Spaces, allowing AI assistants to interact with machine learning models hosted on Hugging Face. Built by Shaun Smith, this Node.js implementation automatically discovers and exposes Gradio endpoints as MCP tools, handling parameter conversion, file uploads, and progress notifications. It supports various output types including text, images, and audio files, with special handling for Claude Desktop compatibility. The server runs locally using stdio transport, making it ideal for workflows requiring access to specialized AI models without leaving the conversation interface.