mcp-hfspace

Hfspace MCP Server

mcp-hfspace connects Claude Desktop to Hugging Face Spaces for image generation, audio tasks, vision analysis, and chat. It serves users who want Space endpoints available through MCP, though the project has been superseded by Hugging Face and Gradio MCP offerings.

Server facts

mcp-hfspace

Tech

Runtime / language
node
Transports
stdio
Remote available
No
Authentication
Not specified

Activity

GitHub stars
389
Last commit
Data as of

Overview

Hugging Face Spaces supply the models behind mcp-hfspace’s image, audio, vision, and conversational workflows. Users configure a selection of Spaces, and the server chooses an appropriate endpoint unless a specific endpoint is supplied. Its default connection uses FLUX.1-schnell for image generation. The exposed tools include FLUX_1-schnell-infer, semantic endpoint discovery through search-spaces, and available-files for identifying saved resources. File listings include resource identifiers, names, sizes, modification times, and media types, helping users distinguish an intended image or audio file. Search results support user selection before further tool calls.

File handling differs by client mode: desktop operation saves generated audio in the working directory and notifies Claude, while operation outside that mode returns embedded resources encoded in base64. A Hugging Face token enables access to private Spaces. The project runs on Node.js over stdio and carries an MIT license. Its documentation now directs users toward replacement MCP services from Hugging Face and Gradio. Documented constraints include unsupported unnamed endpoint parameters, incomplete conversion of complex Python types, and possible timeouts during heavier ZeroGPU jobs.

mcp-hfspace setup notes

  • mcp-hfspace records stdio only; confirm a network transport upstream before planning remote access.

Related MCP servers