How is this different from connecting Hugging Face MCP directly?

Connected directly, an agent sees every tool the server exposes and holds the credential itself. Through TrustGate you choose which tools each agent gets, and every call is logged against the agent that made it.

What if the Hugging Face MCP API changes?

We track the Hugging Face MCP schema, so the tool catalogue stays current. Tools you have already allowed keep working, and anything new stays denied until you enable it.

Can I give different agents different Hugging Face tools?

Yes. Access is set per agent: one can get a handful of read tools, another the full set. You grant the whole MCP or single tools within it.

Where can I see what my agents did in Hugging Face?

Every call is logged against the agent that made it, with the tool it used. Activity can be sent on to your own monitoring.

Do I need to run the Hugging Face MCP server myself?

No. TrustGate connects to the Hugging Face MCP server and sits in front of it. Your agent only talks to your TrustGate URL, never to Hugging Face directly.

Connect Hugging Face MCP to your AI agents, securely.

Q: Is Hugging Face MCP secure?

The server itself exposes 10 tools, and some of them can write to your Hugging Face account. Behind TrustGate an agent only sees the tools you allow, requests and responses are inspected, and every call is logged. Anything outside the allowlist is denied before it reaches Hugging Face.

One secured endpoint for every Hugging Face call your AI agents make, scoped per user, inspected, and logged.

Get started for free

Get a demo

Try Hugging Face MCP now

What your agents can ask once Hugging Face MCP is connected.

“Find the best model for code generation”
“Search for sentiment analysis datasets in English”
“What are the trending models this week?”

“Run this training script as a Hugging Face Job”
“Show me the README for this model”
“Find a Space for background removal”

Select the tools your AI agents can call

10 tools from the Hugging Face MCP schema.

Showing 10 tools

hub_repo_searchreadSearches Hugging Face repositories with a shared query interface, targeting models, datasets, Spaces or all of them.
hf_fsreadBrowses the Hugging Face Hub as a file system: models, datasets, Spaces, collections, papers, trending content, documentation and repository files.
hub_repo_detailsreadGets details for one or more Hugging Face repos, auto-detecting model, dataset or Space, optionally including the README.
create_repowriteCreates a Hugging Face model, dataset, Space or bucket repository.
hf_whoamireadIdentifies the authenticated Hugging Face account, or reports that tools are being used anonymously.
dynamic_spacehigh riskDiscovers and runs Gradio Spaces as tools. Spaces execute third-party code on the Hugging Face infrastructure.
hf_sandboxhigh riskCreates and manages Hugging Face Sandboxes.
hf_sandbox_exechigh riskRuns a shell command inside a Hugging Face Sandbox.
hf_sandbox_fshigh riskReads, writes and manages files inside a Hugging Face Sandbox.
hf_jobshigh riskRuns Python/UV or Docker jobs on Hugging Face remote compute, with submit, logs, inspect, cancel and schedule operations.

What Hugging Face MCP exposes

Hugging Face MCP groups its 10 tools into 4 categories, each with a suggested starting policy.

Category	Covers	Access	Suggested policy
Search	Model, dataset and Space search, plus a shared repo query interface	Read	Allow public hub searches
Repositories	Details for one or more models, datasets or Spaces, including READMEs	Read / Write	Allow repo lookups
Spaces & Account	Running Gradio Spaces as tools and identifying the authenticated account	Read / Write	Approve dynamic Space execution; Spaces run third-party code
Compute	Hugging Face Jobs and Sandboxes: remote job runs, shell commands and sandbox files	Read / Write	Require approval; jobs and sandboxes run arbitrary code and can incur cost