Labsco
MCP SERVER · OFFICIAL PROJECT

Hugging Face Hub Search

by Hugging Face

Four tools over the Hub: search models, datasets and spaces, read repo details, and run filesystem operations on a repo.

Model Routing, Multi-Model Consultation & Cost ControlVerified
Summary
hf_fs batching operations into one call is the design that keeps repo exploration from eating a conversation.

Walking a model repository one file at a time costs a turn per file. Accepting an array of commands means listing, reading and checking can happen in a single round trip — the same instinct behind the batch tools elsewhere here, applied to the place it matters most. hf_whoami is the honest companion: an agent can find out what it is actually allowed to see before it concludes a repository does not exist.

What it is

Hugging Face's endpoint, which listed four tools to our unauthenticated probe. The last one is unusual — a batched filesystem interface over Hub repositories rather than a per-file tool.

What you get
  • hub_repo_search — one query interface across models, datasets, spaces and agents
  • hub_repo_details — details for one or more repositories, with the type auto-detected
  • hf_fs — Hub filesystem operations, taking an array of {cmd, args} items so several run in one call
  • hf_whoami — the current authentication context, including which organisations are visible
Requirements

Nothing for public content. Authenticating changes what hf_whoami reports and what the other tools can see.

Setup effort

Paste a URL — add the endpoint to your client