Skip to content
Get packs

Hugging Face Hub via MCP

Lets your AI assistant search Hugging Face models, datasets, papers and Spaces, and run cloud GPU jobs.

Source Download skill
Repository
huggingface/skills
Stars
11,253
Licence
Apache-2.0
Files
1 file

What it does

Once installed, the agent can query the Hugging Face Hub through an MCP server (a standard connector that gives an AI assistant extra tools). It can search for models, datasets, Spaces and research papers, read model cards and library documentation, call Gradio Spaces as tools (for example image generation or speech-to-text), and launch compute jobs on CPU or GPU hardware with status and log checks. It does not run locally on its own; it only works when the Hugging Face MCP server is connected to your assistant.

Great for

  • ML engineers who want to compare open models and read their documentation without leaving the chat
  • Researchers looking for recent papers and the models linked to them
  • Developers who want to run a training or inference script on a rented cloud GPU
  • Students exploring what open models and datasets exist for a task

Skip it if

  • Anyone who does not want their Hugging Face account token used by an agent: job runs can receive HF_TOKEN secrets and can spend paid compute credits
  • Tasks that must stay fully offline or private, since the skill queries the live Hub and remote Spaces over the network
  • Users who only need general coding help; this skill is specific to the Hugging Face ecosystem

Try asking

  • “Find the best open text-generation model for code and compare the top two options”
  • “Search for English sentiment-analysis datasets and show their model cards”
  • “Run this Python script on a T4 GPU and report whether CUDA is available”
  • “Check the status and logs of my training job that is running on Hugging Face”
Step-by-step guide: how to use Hugging Face Hub via MCPInstall, first run, examples, tips and FAQs

Install in 3 steps

  1. 1

    Download the skill

    The ZIP contains the 1 original files, the Apache-2.0 licence and an attribution note.

    hf-mcp.zip
  2. 2

    Add it to your agent

    Claude Code (all projects)

    mkdir -p ~/.claude/skills && unzip hf-mcp.zip -d ~/.claude/skills/

    Claude Code / Agent SDK (this project only)

    mkdir -p .claude/skills && unzip hf-mcp.zip -d .claude/skills/

    Claude apps (web/desktop): Settings → Capabilities → Skills → Upload skill, and choose the ZIP.

  3. 3

    Ask for the task

    No special command needed — the agent reads the skill's description and loads it when your request matches. Try one of the prompts above.

Get better results

  • Set up the MCP server in your Hugging Face settings first, then confirm the connection before asking for searches or jobs.
  • Review any job script and its flavor before running, because GPU jobs can cost money and secrets are passed to the remote container.
  • Ask for include_readme when you want full model documentation rather than a short summary.

What's inside

  • SKILL.md

Redistributed unmodified from huggingface/skills under the Apache-2.0 licence. The assessment was drafted with AI from the skill's files on 2026-10-29. Skills can instruct your agent to run code — read SKILL.md and any scripts before installing.