Integration · Research and analyze

Hugging Face and Meteor AI agents

Connect Hugging Face to a Met: search models, datasets and Spaces on the Hub, check their details and follow daily papers, read only.

Book a demo See every integration

How does Hugging Face connect to Meteor?

With the Hugging Face ready connector, your Met uses Hugging Face's official MCP server with the account the admin authorizes. It searches Hub repositories, whether models, datasets or Spaces, gets the details of one or several, with a dataset's structure and a preview, and browses the Hub for collections, daily papers, trending models, documentation and files. It helps your team or your agency research what exists before building something. The 4 tools are read only: they do not run models, create repositories or upload files. Usage runs on your Hugging Face account.

What it automates

What a Meteor Met does with Hugging Face

Find an open model for a task

When your team needs, for example, to transcribe audio or classify texts, it asks the Met to search the Hub for models or Spaces for that task, and gets the list with a link to each one.

Check a dataset before using it

The Met gets a dataset's detail: its overview, its structure with configs, splits and schema, and a preview of the rows.

Catch up with what is new on the Hub

The Met browses the Hub for the daily papers, the trending models or a collection, and sums them up for your team.

Read a repository's documentation or files

The Met opens the Hugging Face documentation or a repository's files, such as its card, to answer a specific question from the team.

4 tools

Everything your Met can ask Hugging Face for

Each one is a real action in the connector. You can switch off the ones you would rather it never used.

Tool	What it does
`hf_whoami`	Checks which Hugging Face account is connected, its visible organizations and the credential's access, without ever returning the credential.
`hub_repo_search`	Searches Hugging Face repositories: models, datasets or Spaces, or several types at once, with a link to each one.
`hub_repo_details`	Gets the details of one or several repositories; for a dataset, its overview, its structure or a preview.
`hf_fs`	Browses the Hub: models, datasets, Spaces, collections, papers, daily papers, trending models, documentation and repository files.

The limits

What the Hugging Face integration does NOT do

We would rather say it here than have you find out during rollout.

It does not run models, generate images, create repositories or upload files: the server marks the 4 tools as read only.
It does not include the extra tools each account can add from its MCP page on Hugging Face, such as repos, sandboxes, Jobs or community Spaces.
It does not test or compare models: it shows what the Hub publishes about each repository so your team decides which one to use.
It does not pay your Hugging Face plan. Usage runs on your account; Meteor does not resell the service.

Before you start

What you need to connect Hugging Face

1

A Hugging Face account

Your team's or your agency's Hugging Face account. You can create one at huggingface.co if you do not have it yet.
2

Connect it in Met

In Met, open Connections → Hugging Face → Connect. A Hugging Face window opens so you sign in and authorize Met.

Frequently asked questions

Questions about Hugging Face and Meteor

It lets you research which open models, datasets or Spaces exist before building something, without someone on the team browsing the Hub by hand.

Not with the 4 tools on this page, which only read. Spaces tools that generate content are added from each account's MCP page and are not documented here.

Because it only lists what the server returns today, and the connector directory shows other names. When the two sources do not match, the page goes with the server.

You do, with your Hugging Face account. Meteor does not resell Hugging Face or charge for the queries your Met makes there.

Research and analyze