Hugging Face

Agents can search the Hugging Face Hub — finding models, datasets, and papers, and reading their documentation and metadata.

Note: This is discovery, not execution. Finding a model through this connector does not run it, download it into a workspace, deploy it, or give an agent inference against it. Running a model is a separate matter from reading about one.

If you want an agent to use a model, that is a model provider connection — see Anthropic, OpenAI, OpenRouter, or Grok.

Before you connect

For public Hub content you need nothing at all — no account, no key, no sign-in. Pick Use public Hugging Face Hub and you are done.

For private or gated repositories, you need a Hugging Face account that already has access to them. Connect with an account that belongs to your organization; reach follows the account.

Connect Hugging Face

There are two ways in. Choose the one that matches what agents need to see.

Method What you get What you need
Use public Hugging Face Hub Read-only lookup of public models, datasets, repository details, and documentation Nothing — it is anonymous
Sign in with Hugging Face The authorizing account's reach, including private repositories it can open, plus repository and job tools A Hugging Face account
  1. Open Connectors and select Hugging Face.
  2. Read the access line above the main button. It says who the connection belongs to and which agents can use it. To pick a different identity, narrow the agents, or use another method, select Change.
  3. Choose your method:
    • Use public Hugging Face Hub connects straight away. There is nothing to sign in to.
    • Sign in with Hugging Face opens browser sign-in. ThinkingMach registers its client automatically, so there is nothing to configure in a developer console.

The public method is the shortest connector setup in the catalog, which makes it a good one to try first.

Choose access

Two tiers of content, and the difference is worth stating:

Content Who can see it
Public models, datasets, and papers Anyone. Connecting adds no privileged access to these
Private or gated repositories Only if the authorizing account already has access

The public method only offers read tools for the public Hub, so it is about as low-risk as a connector gets. Two things to keep in mind:

  • Hugging Face rate-limits anonymous requests. A busy agent may hit the limit; if that becomes a problem, switch to a signed-in connection.
  • Queries go to Hugging Face as typed. Keep them to public information — no secrets, personal data, private repository names, or proprietary details.

The signed-in method is where care is needed. Because most of the value here is public content, it is one of the lower-risk connectors to give an agent once its writes are switched off. The exception is an account with access to private organization repositories — that account's reach becomes the agent's.

For a signed-in connection, ThinkingMach asks Hugging Face for the read-mcp, read-repos, contribute-repos, and jobs scopes. That covers reading, but also contributing to repositories and running jobs, so this is not a read-only connection by default. Read the action list on the Permissions tab and set anything that is not a read to Ask first or Off until you need it. See Set action permissions.

Note: A signed-in connection targets the Hub's own tools, and ThinkingMach requests the endpoint with Gradio tooling switched off — so the interactive applications hosted on Spaces are not part of this connection's surface.

Try it

Search Hugging Face for recent text embedding models under 500M parameters and summarize the top three, including their licences.

Expect results you can check on the Hub. Licence is a useful thing to ask for, because it is exactly the detail people skip when picking a model.

Note: Illustrative task, not a recorded test result.

Troubleshooting and limitations

Problem Likely cause Fix
A private repository is not found The authorizing account does not have access Grant access on the Hub, or connect with an account that has it
A gated model's details are unavailable Gated repositories require accepting terms on the Hub Accept the terms with the authorizing account
An agent claims it can run a model it found It cannot — this connector is discovery only Use a model provider connection
A public connection starts failing during heavy use Hugging Face rate-limits anonymous requests Wait and retry, or connect with Sign in with Hugging Face
A private repository is not found on a public connection The public method only sees public content Use Sign in with Hugging Face with an account that has access
Results seem stale The Hub changes constantly; the agent read what was returned at the time Re-run the search
Needs attention The grant was revoked Select Reconnect

Limitations: discovery and metadata only. No inference, no hosting, no Spaces applications. Private content needs a signed-in connection whose account already has access; the public method is anonymous, read-only, and rate-limited.