Provider REST API

15 actions

Connect Claude to Hugging Face

Search and inspect models, datasets, Spaces, papers and collections on the Hugging Face Hub, and create or manage your own repositories. Toolspoke puts 15 of its actions behind one MCP endpoint that Claude, Cursor, and Codex all speak.

Connect View pricing

Connection: Provider REST API
Authentication: Access token
Actions exposed: 15
What a call bills: Nothing
Adapter: Maintained by Toolspoke

Connected in three steps

1
Install Hugging Face
Open the marketplace in your workspace, add Hugging Face to the project your agents work in, and it appears on the gateway immediately.
2
Connect the credential
Authenticate with access token. Where to get one, and what it has to be able to reach, is the next section.
3
Point your agent at the gateway
Give your client one address, https://toolspoke.com/mcp. Claude Code takes it as a command, Claude and Claude Desktop add it as a custom connector, and Cursor, Codex and VS Code each read it from a config file of their own.

.mcp.json

{
  "mcpServers": {
    "toolspoke": {
      "type": "http",
      "url": "https://toolspoke.com/mcp"
    }
  }
}

One block covers every tool you have installed. Hugging Face shows up in the client as soon as your policy allows it, and so does everything else you install later.

Where the address goes, per client

Claude Code: Run it in your project, then /mcp to sign in
claude mcp add --transport http toolspoke https://toolspoke.com/mcp
Claude and Claude Desktop: Settings, then Connectors, then Add custom connector
https://toolspoke.com/mcp
Cursor: ~/.cursor/mcp.json, or .cursor/mcp.json for one project
{ "mcpServers": { "toolspoke": { "url": "https://toolspoke.com/mcp" } } }
Codex: ~/.codex/config.toml
[mcp_servers.toolspoke] url = "https://toolspoke.com/mcp"
VS Code: .vscode/mcp.json, or the MCP: Add Server command
{ "servers": { "toolspoke": { "type": "http", "url": "https://toolspoke.com/mcp" } } }

What Hugging Face asks for

Access token. You provide it once, when you install the connector. Toolspoke encrypts it at rest and decrypts it only for the length of a single call, and the gateway attaches it to the outbound request itself, so it is never part of the arguments an agent sends.

Access tokenRequired: huggingface.co → Settings → Access Tokens → Create new token. A Read token covers every read here; create_repo, update_repo_settings and delete_repo need a Write token, or a fine-grained one with write access to the namespaces you will change.; hf_…

What Claude can do in Hugging Face

15 actions, each one declared and named by the connector rather than discovered at runtime. A workspace policy grants a person all of them, a hand-picked selection, everything on the read side, everything on the write side, or none.

Reads: 12Reads
Writes: 1Writes
Destructive: 2Destructive

Reads

Fetches data and changes nothing.

whoami
Return the account this token belongs to, its organizations, and the token's role and fine-grained scopes. Costs nothing, and it is the only call that fails on a bad token - every public read would succeed without one - so it is what an install tests itself with.
search_models
Search models on the Hub by keyword, author, tag or task, sorted descending by the chosen field. Returns ids, download and like counts, tags and pipeline task.
get_model
Retrieve one model's metadata - tags, pipeline task, library, card data, files, downloads and likes.
search_datasets
Search datasets on the Hub by keyword, author or tag, sorted descending by the chosen field.
get_dataset
Retrieve one dataset's metadata - tags, card data, files, downloads and likes.
search_spaces
Search Spaces (hosted demo apps) by keyword, author or tag, sorted descending by the chosen field.
get_space
Retrieve one Space's metadata - SDK, hardware, runtime stage, tags and likes.
list_repo_files
List the files at the root of a model, dataset or Space at one revision, with sizes and LFS info. Set recursive to walk the whole tree.
list_commits
List the commit history of a model, dataset or Space at one revision, newest first.
search_papers
Search research papers indexed on Hugging Face Papers by text. Returns titles, arXiv ids, abstracts and upvotes.
list_daily_papers
List the Daily Papers curated on Hugging Face, for today or one date.
list_collections
Search curated collections of models, datasets, Spaces and papers, by owner, text or a repository they contain.

Writes

Creates or updates something on the other side.

create_repo
Create an empty model, dataset or Space repository under your account or an organization. Created private unless private is set to false. A Space also needs an sdk.

Destructive

Deletes or permanently alters something. Worth granting on purpose.

update_repo_settings
Change a repository's visibility or access gating. Destructive because making a private repository public publishes its files and history to everyone, and that cannot be taken back once they are downloaded.
delete_repo
Delete a model, dataset or Space repository permanently, with every file, commit and discussion in it. There is no undo.

What it will not do

Enforced by the gateway rather than left to convention, which is why each of these can be stated flatly.

It cannot call anything else: The 15 actions above are the whole of it. A call to any other name is refused before it reaches Hugging Face rather than forwarded on, and connecting your account does not add to the list: it is fixed by the connector, not discovered at run time.
It reaches no further than your credential: Toolspoke holds no access to Hugging Face of its own. Every call carries the credential you stored and nothing besides, so whatever that credential cannot reach, this connector cannot reach either.
It never hears from Hugging Face: Nothing is pushed to it. There is no webhook, no subscription and no polling, so this connector cannot notice by itself that something changed in Hugging Face. An agent has to ask.
It does not smooth over provider limits: Toolspoke does not retry, queue or back off around Hugging Face's own rate limits. A call that Hugging Face refuses comes back to the agent as a failed call.

Before you connect it

What can Claude do in Hugging Face?

15 named actions: 12 that only read, 1 that write and 2 that delete or permanently alter something. They include whoami, search_models and get_model. Nothing outside that list is reachable: the connector declares each operation by name rather than proxying whatever an agent asks for.

What credentials does the Hugging Face connector need?

Access token. The connector asks for access token. Values are encrypted at rest and attached to the outbound request by the gateway, so they are never part of the arguments an agent sends and never reach the audit log.

Does the Hugging Face connector work with Cursor and Codex, or only Claude?

Any client that speaks MCP, and every one of them gets the same 15 actions. There is a single address, https://toolspoke.com/mcp. Claude Code adds it with claude mcp add --transport http, Claude and Claude Desktop take it as a custom connector in settings, Cursor reads it from .cursor/mcp.json, Codex from ~/.codex/config.toml, and VS Code from .vscode/mcp.json. Each of them signs in to the gateway itself, so there is no key to paste.

What does the Hugging Face connector not do?

The 15 actions above are the whole of it. A call to any other name is refused before it reaches Hugging Face rather than forwarded on, and connecting your account does not add to the list: it is fixed by the connector, not discovered at run time. Toolspoke holds no access to Hugging Face of its own. Every call carries the credential you stored and nothing besides, so whatever that credential cannot reach, this connector cannot reach either. Nothing is pushed to it. There is no webhook, no subscription and no polling, so this connector cannot notice by itself that something changed in Hugging Face. An agent has to ask. Toolspoke does not retry, queue or back off around Hugging Face's own rate limits. A call that Hugging Face refuses comes back to the agent as a failed call.

Can I limit which actions an agent can call?

Yes, in two places. The project switches Hugging Face's actions on and off one at a time, for everyone in the project at once, and the screen groups them by read, write and destructive so turning off everything that deletes is one click. An individual agent key can then be narrowed further, to particular toolkits in a project and to particular actions in a toolkit. Whatever it was granted, a key never reaches a project its owner cannot.

What gets recorded when an agent calls Hugging Face?

Every attempt, with the agent that made it and the person that agent belongs to, the full request payload, the response payload, the status, the duration, and what the call cost in sandbox time and model tokens. Values whose key names a secret are masked out before the record is shown to anyone. An operation the connector marks as not retained never has its response body written at all, so the gateway keeps no second copy of what was read.

More ai connectors

Browse all of them