Skip to content
Kitbay
Submit

Hugging Face MCP servers and agent skills

8 MCP servers and 69 agent skills connect AI assistants to Hugging Face. Includes Hugging Face's official listings.

Platform for sharing machine learning models, datasets and demo apps.

8 MCP servers69 skills26 officialHugging Face

Last checked

All listings

77

Sorted: official first, then quality and activity.Show all, including less active listings

How AI connects to Hugging Face

The huggingface_hub Python library works with the Hugging Face Hub: downloading and uploading files, creating and managing repositories, joining discussions and running inference. Hugging Face also documents its own MCP server, configured from the user's MCP settings page on the Hub, which lets AI assistants search models, datasets, Spaces, papers and documentation, run jobs and sandboxes, and call MCP-compatible Gradio Spaces as tools.

From the vendor's documentation: Hub client library (huggingface_hub), Hugging Face MCP Server

Guides:MCP servers vs agent skillsConnecting MCP servers to AI clients

Questions

Is there an official Hugging Face MCP server?
Yes. 26 listings for Hugging Face are published by the tool's vendor, marked Official in the table above with a link to the evidence.
Can Claude use Hugging Face?
Yes, through an MCP server. 8 MCP servers connect Claude and other AI assistants to Hugging Face; each listing shows how it connects and how to install it.
Are there agent skills for Hugging Face?
Yes: 69 skills are listed. Skills are folders of instructions (and sometimes scripts) that an agent loads when a task needs them.
Which AI clients do Hugging Face servers and skills support?
The listings on this page most often document claude-code, cursor, gemini-cli, codex. Each listing shows the clients its own documentation names.