hf-mcp

Search and retrieve Hugging Face Hub models, datasets, and Spaces via MCP server.

10.9k|724|Updated Nov 24, 2025
One-click install
npx skills add https://github.com/huggingface/skills --skill hf-mcp
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: hf-mcp
Source: https://github.com/huggingface/skills/tree/main/hf-mcp/skills/hf-mcp
Command: npx skills add https://github.com/huggingface/skills --skill hf-mcp

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This skill enables AI assistants to access Hugging Face Hub resources through MCP server to search and retrieve models, datasets, Spaces, papers, and documentation.

Core Features & Use Cases

  • Model, Dataset, and Space Discovery: Search and retrieve models, datasets, and Spaces through MCP-enabled tooling, including hub_repo_details and space exploration.
  • Documentation & Compute: Fetch model/docs details and run compute jobs or tool-based Spaces, enabling integrated AI workflows.
  • Use Case: When connected to an MCP server, an agent can locate the best model for a task, fetch relevant docs, and launch a Gradio Space as a tool.

Quick Start

Use the hf-mcp skill to connect to the MCP server and run a sample search for models.

Frequently Asked Questions about hf-mcp

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I search for Hugging Face models and datasets using an AI assistant?

You can search Hugging Face models and datasets by connecting an AI assistant to an MCP server with model_search and dataset_search commands to discover and retrieve Hub resources directly.

Can I fetch Hugging Face documentation through an MCP server?

Yes, you can retrieve Hugging Face documentation through an MCP server using the hf_doc_search command, allowing AI assistants to fetch relevant docs for integrated workflows.

What do I need to use MCP for Hugging Face Hub discovery?

You need MCP client tooling available in your workspace to use this skill for Hugging Face Hub discovery, enabling tool-based interactions like model_search and hub_repo_details.

Can an AI agent launch Hugging Face Spaces or compute jobs via MCP?

Yes, an AI agent can launch Gradio Spaces as tools and run compute jobs using the dynamic_space and hf_jobs commands when connected to an MCP server.

What Hugging Face Hub resources can I retrieve through an MCP server?

You can retrieve models, datasets, Spaces, papers, and documentation through an MCP server, supporting comprehensive Hub resource discovery and repository details retrieval.

How does MCP integration work for discovering Hugging Face Spaces?

MCP integration enables AI assistants to use the space_search command to explore and retrieve Hugging Face Spaces, providing hub_repo_details and tool-based Space interactions.