Hugging Face MCP Server

Hugging Face Official MCP Server

Getting Started

This repository contains the offical Hugging Face MCP Server source. The server supports multiple deployment and configuration options, and is deployable locally as well as on your own infrastructure.

The easiest way to access Hugging Face MCP Services is via https://hf.co/mcp which offers a number of Tools, Prompts and access to Gradio Spaces which are configurable at https://hf.co/settings/mcp.

Anonymous access is supported with a default set of Tools and Prompts, or use a Hugging Face READ token to customize your settings:

If you prefer to use OAuth, use https://hf.co/mcp?login

For claude.ai enter https://hf.co/mcp from the "Add Integrations" dropdown menu. Claude does not currently support the MCP 2025-06-18 OAuth standard so is limited to anonymous access.

Quick Guide (Repository Packages)

This repo contains:

• (/mcp) MCP Implementations of Hub API and Search endpoints for integration with MCP Servers.
• (/app) An MCP Server and Web Application for deploying endpoints.

MCP Server

The following transports are supported:

• STDIO
• SSE (To be deprecated, but still commonly deployed).
• StreamableHTTP
• StreamableHTTP in Stateless JSON Mode (StreamableHTTPJson)

The Web Application and HTTP Transports start by default on Port 3000.

SSE and StreamableHTTP services are available at /sse and /mcp respectively. Although though not strictly enforced by the specification this is common convention.

Running Locally

You can run the MCP Server locally with either npx or docker.

To run with docker:

All commands above start the Management Web interface on http://localhost:3000/. The Streamable HTTP server is accessible on http://localhost:3000/mcp. See [Environment Variables](#Environment Variables) for configuration options. Docker defaults to Streamable HTTP (JSON RPC) mode.

Development

This project uses pnpm for build and development. Corepack is used to ensure everyone uses the same pnpm version (10.12.3).

Build Commands

pnpm run clean -> clean build artifacts

pnpm run build -> build packages

pnpm run start -> start the mcp server application

pnpm run buildrun -> clean, build and start

pnpm run dev -> concurrently watch mcp and start dev server with HMR

Docker Build

Build the image:

Run with default settings (Streaming HTTP JSON Mode), Dashboard on Port 3000:

Run STDIO MCP Server:

TRANSPORT can be stdio, sse, streamingHttp or streamingHttpJson (default).

Transport Endpoints

The different transport types use the following endpoints:

• SSE: /sse (with message endpoint at /message)
• Streamable HTTP: /mcp (regular or JSON mode)
• STDIO: Uses stdin/stdout directly, no HTTP endpoint

Stateful Connection Management

The sse and streamingHttp transports are both stateful - they maintain a connection with the MCP Client through an SSE connection. When using these transports, the following configuration options take effect:

Environment Variables

The server respects the following environment variables:

• TRANSPORT: The transport type to use (stdio, sse, streamableHttp, or streamableHttpJson)
• DEFAULT_HF_TOKEN: ⚠️ Requests are serviced with the HF_TOKEN received in the Authorization: Bearer header. The DEFAULT_HF_TOKEN is used if no header was sent. Only set this in Development / Test environments or for local STDIO Deployments. ⚠️
• If running with stdio transport, HF_TOKEN is used if DEFAULT_HF_TOKEN is not set.
• HF_API_TIMEOUT: Timeout for Hugging Face API requests in milliseconds (default: 12500ms / 12.5 seconds)
• USER_CONFIG_API: URL to use for User settings (defaults to Local front-end)
• MCP_STRICT_COMPLIANCE: set to True for GET 405 rejects in JSON Mode (default serves a welcome page).
• AUTHENTICATE_TOOL: whether to include an Authenticate tool to issue an OAuth challenge when called
• SEARCH_ENABLES_FETCH: When set to true, automatically enables the hf_doc_fetch tool whenever hf_doc_search is enabled