OpenBot
DownloadApp
All plugins

Hugging Face

Models, datasets and papers on the Hub

Hugging Face lets agents search models, datasets, Spaces, collections and papers on the Hugging Face Hub, read model cards and repository files, preview datasets, and search the Hugging Face documentation. It uses public Hub data and needs no account and no key. The tools only read: agents cannot create repositories or run jobs.

What an agent can do with Hugging Face

  • Find models, datasets and Spaces

    The hub_repo_search tool searches models, datasets and Spaces on the Hugging Face Hub, one type or all types in one call. The hf_fs tool can also filter by Hub tag, for example to find the fine-tunes or quantized versions of a base model.

  • Inspect a repository or a dataset

    The hub_repo_details tool gets the details of a model, dataset or Space. For a dataset, it shows the configs, splits, sizes and schema, and a preview of the rows. The hf_fs tool reads files from a repository, such as a model card.

  • Follow papers and trends

    The hf_fs tool lists trending models, datasets, Spaces and papers, and the latest daily papers. It reads the text and metadata of a paper by its arXiv ID. The agent can tell you which new papers relate to your task.

  • Search the documentation

    The hf_fs tool searches the Hugging Face documentation with a natural language query. Ask, for example, how to use LoRA adapters with PEFT, and the agent gets the matching pages.

How to connect Hugging Face to OpenBot

  1. Open Hugging Face in OpenBot

    Press Open in OpenBot on this page, or find Hugging Face in the app's Apps tab. Get the app first if it is not on this computer.

  2. Give an agent a task

    OpenBot asks for no sign-in or key for Hugging Face. Ask an agent for work that needs Hugging Face, such as one of the example prompts below. The agent uses the plugin's tools to do it.

Example prompts3

  • Hugging FaceFind small open models for speech recognition on Hugging Face.
  • Hugging FaceWhich papers are trending on Hugging Face today?
  • Hugging FaceFind a dataset of weather time series and show its columns.

Apps1

  • Hugging Face

    Model, dataset, Space, paper, and docs search, over Hugging Face's MCP server with no sign-in.

Information6

Developer
Hugging Face
Category
Research
Version
1.0.0
Website
huggingface.co
Privacy policy
huggingface.co/privacy
Terms
huggingface.co/terms-of-service

Questions

Is the Hugging Face MCP server official?

Yes. Hugging Face hosts the MCP server at huggingface.co/mcp and documents it in the Hub documentation. The source code is in Hugging Face's hf-mcp-server repository on GitHub.

Does the Hugging Face plugin need an account or a key?

No. The OpenBot listing connects with no sign-in. Hugging Face answers requests without a token with its anonymous set of tools. The server tells the agent that anonymous use has rate limits. A Hugging Face account gives higher limits.

What can the agent see and change on Hugging Face?

The agent reads public Hub data: models, datasets, Spaces, collections, papers and documentation. The anonymous tools only read, so the agent cannot change anything on the Hub. Hugging Face gets the requests that the agent sends. Your workspaces, conversations and files stay on the computer that runs OpenBot.

Can the agent create repositories or run jobs?

Not with this listing. Hugging Face says that repository creation, Jobs, dynamic Spaces and sandbox tools need Hugging Face authentication. You turn on these tools on your MCP settings page on Hugging Face.

Sources

Every statement about Hugging Face comes from the pages below, checked on . They can change after that date.