Hf MCP Server
Hugging Face Hub and Gradio app access for your AI assistant workflows.
Data last scanned today · Reviewed today
Overview
This server integrates the Hugging Face ecosystem directly into your AI assistant, providing tools to interact with the Hub and various Gradio applications. It is designed for use within environments like Claude Desktop, Gemini CLI, VSCode, and Cursor. After installation, users configure their specific tools and accessible Spaces through the Hugging Face settings page. The server supports authentication via Bearer tokens for tailored access.
Our verdict
This server holds a high risk level with a security score of 48/100, primarily because the authentication method and read-only mode have not been independently reviewed. While it is actively maintained with 15 contributors and zero direct dependencies, potential users should exercise caution as it is community-maintained rather than developed by the official vendor.
- Zero direct dependencies reduce supply-chain surface area
- Active development with 15 contributors
- Supports multiple AI IDE and CLI environments
- Authentication method lacks an independent security review
- Read-only mode implementation has not been reviewed
- Operates as a community-maintained project
Setup
{
"mcpServers": {
"hf-mcp-server": {
"args": [
"-y",
"hf-mcp"
],
"command": "npx"
}
}
}Paste into ~/Library/Application Support/Claude/claude_desktop_config.json (macOS) or %APPDATA%\Claude\claude_desktop_config.json (Windows), then restart Claude Desktop. Some servers require API keys or tokens as environment variables — check the project's README and add an "env" object if needed.
- No read-only mode — running this grants full read/write access.
- No built-in authentication — anyone with access to this config can invoke every tool it exposes.
Compatibility
| Client | Local | Docker | Remote | Read-only |
|---|---|---|---|---|
| Claude Desktop | ||||
| Cursor | ||||
| VS Code | ||||
| Windsurf | ||||
| ChatGPT |
Frequently asked questions
›How do I configure which Tools and Spaces are available?
Once the server is installed, navigate to https://huggingface.co/settings/mcp to manage your Tools and Spaces configuration.
›Can I remove image content from Gradio server responses?
Yes, you can append ?no_image_content=true to the server URL to prevent ImageContent blocks from being included in the responses.
›Does the server require an authentication token?
Yes, when using HTTP, you must provide your Hugging Face token in the Authorization: Bearer header. Some installation methods allow you to follow an interactive login prompt instead.
›What transport protocols are supported?
The server supports both STDIO and StreamableHTTP in Stateless JSON mode.
›Where is the management interface located when running locally?
The management web interface runs on http://localhost:3000/ by default, with the Streamable HTTP service available at /mcp.
Changelog
Badge
Maintain this server? Add the live badge to your README.