← Explore all mcp servers

ML RESOURCE DISCOVERY / Hugging Face

Hugging Face MCP

Connects an assistant to Hub discovery, with optional community tools and account-scoped compute or repository actions.

“Search and explore Hub resources: models, datasets, Spaces, papers and more.”

01 / THE REASONING

Why this made the selection.

  • Combines specialist model, dataset, paper and documentation discovery in an existing Hub workflow.
  • The MCP settings distinguish extra tools and community Spaces, so users can inspect the requested capabilities before enabling them.

02 / THE REVIEW RECORD

What we actually inspected.

Source review has boundaries.
A clear record is more useful than a “safe” badge.

Material inspected

  • Official MCP guide: current built-in tools, settings and optional capabilities.
  • Agents overview: Claude Code connection command.
  • Spaces MCP guide and Jobs billing documentation: token, quota and compute implications.
  • Official registry entry co.huggingface/hf-mcp-server: remote transport and endpoint.

Our findings

  • The current guide names hf_fs for Hub navigation; optional repository, Sandbox and Jobs tools expand account authority.
  • Space tools can consume the user’s compute quota, while Jobs have separate hardware-usage billing.

Not established by this review

  • Hosted implementation, authentication, community tools, compute execution and data retention were not tested.

The review applies to the material and revision named here. A newer upstream release can change its behavior.

03 / PUT IT TO WORK

Connect a server deliberately.

Upstream setup instructions ↗

DOCUMENTED COMMAND

claude mcp add hf-mcp-server -t http "https://huggingface.co/mcp?login"

Copying does not execute this command. It may retrieve a newer version than the reviewed source.

  1. For Claude Code use the documented HTTP command; for other clients copy the generated configuration from your Hugging Face MCP settings.
  2. Select the required tools and Spaces deliberately, then reload the client.

Before you start

  • A Hugging Face account for personalized setup and the selected MCP client.
  • Appropriate token/account permissions and compute access for the tools you enable.

04 / FOLLOW THE EVIDENCE

The source trail.

Our notes are separate from the original resource.
Check upstream before adopting a new version.

  1. https://huggingface.co/docs/hub/agents-mcp

    Supports: upstreamDescription, descriptionSourceUrl, summary, whySelected, bestFor, limitations, install, compatibility, review, access.permissions

  2. https://huggingface.co/docs/hub/agents-overview

    Supports: install, compatibility

  3. https://huggingface.co/docs/hub/spaces-mcp-servers

    Supports: limitations, install.prerequisites, review, access

  4. Jobs billing ↗Checked

    https://huggingface.co/docs/hub/jobs-pricing

    Supports: review, access.cost

  5. https://registry.modelcontextprotocol.io/v0.1/servers?search=huggingface&limit=10

    Supports: transports