> ## Documentation Index
> Fetch the complete documentation index at: https://docs.atako.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Hugging Face

> Connect Hugging Face to your Atako agents — 14 read and 1 write actions.

Let your agents search and read models, datasets, Spaces and collections on the Hugging Face Hub, browse repository files and take part in repository discussions.

## Connection

* **Authentication**: API key (User access token).

<Note>
  huggingface.co → Settings → Access Tokens → New token. Name it and pick a role: "Read" covers every read action; "Write" (or a "Fine-grained" token allowed to interact with discussions) is required for create\_discussion. Copy the token (starts with hf\_).

  See [Hugging Face's documentation](https://huggingface.co/settings/tokens).
</Note>

## Read actions (14)

| Action | Description |
| - | - |
| `get_collection` | Get a collection and its items. collection\_slug (string "namespace/title-id", the `slug` returned by list\_collections). |
| `get_dataset` | Get a dataset's metadata: description, tags, downloads, likes, files list, card data. repo\_id (string "namespace/name", the `id` returned by the search actions). |
| `get_discussion` | Get one discussion or pull request with its comments and status events. repo\_type ("models" \| "datasets" \| "spaces"); repo\_id (string "namespace/name", the `id` returned by the search actions); num (integer, the discussion number from list\_discussions). |
| `get_model` | Get a model card's metadata: task, library, tags, downloads, likes, files list, config. repo\_id (string "namespace/name", the `id` returned by the search actions). |
| `get_space` | Get a Space's metadata: SDK, runtime stage and hardware, tags, likes, files list. repo\_id (string "namespace/name", the `id` returned by the search actions). |
| `list_collections` | List collections, filtered by owner (user or organization name), by an item they contain (item: "models/namespace/name", "datasets/…", "spaces/…" or "papers/…") or by text (q), sorted by upvotes, lastModified or trending. Each result has a slug for get\_collection. |
| `list_discussions` | List discussions and pull requests of a repository. repo\_type ("models" \| "datasets" \| "spaces"); repo\_id (string "namespace/name", the `id` returned by the search actions); type ("all" \| "discussion" \| "pull\_request"); status ("all" \| "open" \| "closed"); author (username); search (text); sort ("recently-created" \| "trending" \| "reactions"); p (0-based page). |
| `list_my_repositories` | List the repositories of the token owner (id "namespace/name", type, visibility, storage used). Use it to discover your own repo\_id values. |
| `list_notifications` | List the token owner's Hub notifications (discussions, pull requests, papers), 0-based page p, optionally only unread ones or only those for one repository. |
| `list_repo_files` | List files and folders of a repository (metadata only: type, path, size, LFS info; with expand, last commit per entry). repo\_type ("models" \| "datasets" \| "spaces"); repo\_id (string "namespace/name", the `id` returned by the search actions); revision (string: branch, tag or commit sha, e.g. "main" — no slash, so give a pull-request ref as its commit sha); path (optional folder inside the repo, e.g. "onnx"); recursive (boolean); expand (boolean); limit (integer). |
| `search_datasets` | Search datasets on the Hub by text, author or tag (filter, e.g. "task\_categories:text-classification", "language:fr"), sorted by downloads, likes, trendingScore, createdAt or lastModified (descending). |
| `search_models` | Search models on the Hub by text, author, tag (filter, e.g. "text-classification", "pytorch", "license:apache-2.0") or task (pipeline\_tag), sorted by downloads, likes, trendingScore, createdAt or lastModified (descending). |
| `search_spaces` | Search Spaces (hosted demo apps) on the Hub by text, author or tag (filter, e.g. "gradio"), sorted by likes, trendingScore, createdAt or lastModified (descending). |
| `whoami` | Get the account the token belongs to: type, id, name (the namespace of your repositories), fullname, email, emailVerified, isPro, avatarUrl and orgs (name, fullname, roleInOrg). Token details (auth) are removed. No argument. |

## Write actions (1)

| Action | Description |
| - | - |
| `create_discussion` | Open a new discussion on a repository (not a pull request). Arguments: repo\_type (string: "models" \| "datasets" \| "spaces"); repo\_id (string "namespace/name", the `id` returned by the search actions); title (string, 3 to 200 characters); description (string, Markdown body, up to 65536 characters). |

## Permissions

Every action above must be explicitly granted to an agent before it can be used. See [Permissions](/integrations/permissions) for the grant model and [Security](/integrations/security) for how credentials are protected.

***

*Last reviewed against the provider API: September 2026.*


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.