Hugging Face integration

Lets your app search and read models, datasets and Spaces on the Hugging Face Hub, and run chat completions on open models through Inference Providers, billed to your Hugging Face account.

Connecting

On the Integrations page choose Connect on Hugging Face and fill in:

• Access token — huggingface.co → Settings → Access Tokens → Create new token. A fine-grained token with “Make calls to Inference Providers” is enough.

Sapilon checks the key with the provider before it saves it, and stores it encrypted; it is never shown again.

• Inference calls are billed to the Hugging Face account whose token you use.

What your app can do

Only these requests are available; anything else is refused.

• GET /models
• GET /models/**
• GET /datasets
• GET /datasets/**
• GET /spaces
• GET /spaces/**
• POST /v1/chat/completions

How the agent uses it

Use integrations.request('hugging-face', …); paths are relative to https://huggingface.co/api.

• GET /models — Example: { method: 'GET', path: '/models', query: { search: 'whisper', sort: 'downloads', limit: '10' } }. GET /models/{owner}/{name} — one model.
• GET /datasets, GET /datasets/{owner}/{name}; GET /spaces, GET /spaces/{owner}/{name}.
• POST /v1/chat/completions (sent to router.huggingface.co automatically) — OpenAI-style. Example: { method: 'POST', path: '/v1/chat/completions', body: { model: 'meta-llama/Llama-3.3-70B-Instruct', messages: [{ role: 'user', content: 'Hello' }] } }.

For the app’s own AI features prefer the platform’s ai helper; use this for a specific open model. Writing to the Hub is not allowed.

More from Hugging Face: https://huggingface.co/docs/hub/api