Menu
Mattering Automations connector

The Hugging Face MCP for automations that run models and read the Hub

Connect one Hugging Face token and serverless inference and the Hub catalog become steps on a visual canvas that AI agents can drive: chat completions routed across Inference Providers, embeddings, summarization, translation, classification, image generation, and speech transcription, next to search and metadata for nearly three million models, plus datasets and Spaces. 36 typed nodes, one credential.

36 ready-made nodes One credential to connect Works with Claude, ChatGPT, and Matter

What the Hugging Face MCP connector does

Hugging Face is two products behind one token, and the Mattering Automations connector ships both as 36 typed nodes: 22 read-only queries and 14 actions. The inference family rides Inference Providers, Hugging Face's serverless router: an OpenAI-compatible chat completion with server-side provider routing, plus task pipelines on Hugging Face's own hf-inference compute for embeddings, summarization, translation, zero-shot and standard text classification, image generation, speech transcription, and image classification. The Hub family covers the catalog itself: search and inspect models, datasets, and Spaces, walk file trees and refs, follow collections, papers, and trending repos, check who liked what, and create or delete repos.

Every node speaks MCP, the open protocol for connecting AI assistants to tools. Claude, ChatGPT, and any MCP-capable assistant can call them, and so can Matter, the assistant built into the platform. Hugging Face also ships its own official MCP server at hf.co/mcp, which is excellent at what it does; this connector exists for a different job. Here the nodes live on an automation canvas, so a scheduled workflow can summarize the Daily Papers feed every morning or watch a model release with no assistant in the loop, and the same nodes chain with the rest of your stack: the model's answer lands in a task, a message, or a database row in the next step.

Two Hugging Face facts shape the node list. First, the chat node uses the router's OpenAI-compatible endpoint, which Hugging Face documents for chat completion tasks only, so the non-chat work ships as dedicated pipeline nodes pinned to hf-inference, the one provider that serves the classic task pipelines without a third-party account. Second, every model field is free text with a documented default rather than a fixed dropdown, because the Hub hosts around three million models and the served line-up changes constantly; Hugging Face: List Router Models returns every chat model currently served with per-provider pricing and latency, and Hugging Face: Search Models filters the Hub by task, library, or which provider serves a model. Cold pipeline models return a 503 with an estimated wait, and the connector surfaces that as a retry hint rather than letting it read like an auth failure.

One connection, three surfaces

  • Visual automations. Drag Hugging Face nodes onto the canvas and wire them to triggers, conditions, and other apps.
  • AI agents, yours or ours. These nodes speak MCP, the open protocol, so Claude, ChatGPT, and any MCP-capable assistant can use them, and Matter, the agent built into the platform, uses them natively.
  • The whole platform. Combine with native tasks, projects, and other connectors, no middleware in between.

Every Hugging Face node, ready to drop on the canvas

36 nodes, generated from the connector itself so this list always matches what ships. Expand a category to browse.

Read and query nodes 22

Hugging Face: Get Collection

Get one collection with its ordered items (models, datasets, Spaces, and papers) and per-item notes. Free read.

Hugging Face: Get Dataset

Get full metadata for one dataset: downloads, likes, tags, card data, and sibling files. Free read.

Hugging Face: Get Model

Get full metadata for one model: downloads, likes, pipeline tag, tags, license card data, sibling file list, and gating status. Ask for extra computed properties (downloadsAllTime, inference, inferenceProviderMapping) via expand[]. Free read.

Hugging Face: Get Paper

Get one indexed paper by its arXiv id: title, abstract, authors, and community upvotes. Free read.

Hugging Face: Get Router Model

Get the provider line-up for one served chat model: which providers are live, their pricing, context windows, and capability flags. Free read.

Hugging Face: Get Space

Get full metadata for one Space: SDK, likes, tags, sibling files, and its current runtime block (stage and hardware). Free read.

Hugging Face: Get Space Runtime

Get the live runtime state of a Space: stage (RUNNING, SLEEPING, BUILDING, RUNTIME_ERROR, ...), current and requested hardware, and sleep timeout. Handy for monitoring a deployed demo from an automation. Free read.

Hugging Face: Get User Overview

Get a user's public profile overview: full name, plan, and counts of models, datasets, Spaces, followers, and likes. Free read.

Hugging Face: List Collections

List curated Hub collections, optionally filtered to one owner or to collections containing a specific item. Free read.

Hugging Face: List Daily Papers

List the community-curated Daily Papers feed - trending ML research with upvotes and summaries. Great source material for a research digest automation. Free read.

Hugging Face: List Dataset Files

List the files in a dataset repo at a given revision, with size and LFS info per entry. Paginated via cursor. Free read.

Hugging Face: List Model Files

List the files in a model repo at a given revision: path, size, and LFS pointer info per entry. Paginated via cursor. Free read.

Hugging Face: List Model Refs

List the branches and tags of a model repo with the commit each one points at. Free read.

Hugging Face: List Repo Likers

List the users who liked a repo (model, dataset, or Space). Paginated. Free read.

Hugging Face: List Router Models

List every chat-completion model currently served by Inference Providers, with per-provider status, context length, pricing per million tokens, tool support, latency, and throughput. Use it to pick a model before a Chat Completion node. Free read.

Hugging Face: List Trending Repos

List what is trending on the Hub right now across models, datasets, and Spaces. A ready-made source for a weekly AI-news digest automation. Free read.

Hugging Face: List User Likes

List the public repos (models, datasets, Spaces) a user has liked, newest first. Free read.

Hugging Face: Search Datasets

Search and filter the Hub's dataset catalog by keyword, author, or tag. Returns id, downloads, likes, and tags per dataset. Free read.

Hugging Face: Search Models

Search and filter the Hub's model catalog: by keyword, author, task (pipeline_tag), library, or which inference provider currently serves the model. Returns id, downloads, likes, and tags per model. Free read.

Hugging Face: Search Papers

Full-text search over papers indexed on the Hub. Returns matching papers with ids and summaries; feed an id into the Get Paper node for detail. Free read.

Hugging Face: Search Spaces

Search and filter Hugging Face Spaces (hosted ML demo apps) by keyword, author, or tag. Returns id, likes, and SDK per Space. Free read.

Hugging Face: Who Am I

Validate the connected Hugging Face token and return the account behind it: username, orgs, plan, and the token's own role and scopes. Free and side-effect free, so it doubles as the credential health check.

Action nodes 14

Hugging Face: Chat Completion

Run an OpenAI-compatible chat completion against any conversational model served by Hugging Face Inference Providers. The router picks the fastest live provider by default; steer it by appending :cheapest or a provider name like :groq to the model id. The reply text lands at data.choices[0].message.content. Consumes Inference Providers credits.

Hugging Face: Classify Image

Classify an image with a vision model: send the image as a base64-encoded string and get label/score pairs back. Useful for content screening or routing uploads in an automation.

Hugging Face: Classify Text

Classify text with a model trained for a fixed label set - sentiment, toxicity, language detection, and similar. Returns label/score pairs. For labels the model was not trained on, use the Zero-Shot node instead.

Hugging Face: Create Embeddings

Turn text into embedding vectors (feature extraction) with a sentence-transformers style model. Send one string or a JSON array of strings and get one float vector per input as an array of arrays. Keep batches modest: every vector is a large float array in the execution record.

Hugging Face: Create Repo

Create a new empty repo on the Hub: a model, dataset, or Space (Spaces also need an sdk). The token must have write access. Reversible with the Delete Repo node.

Hugging Face: Delete Repo

Permanently delete a repo (model, dataset, or Space) the token can administer. This cannot be undone and removes all files and history - point it only at repos this automation created.

Hugging Face: Generate Image

Generate an image from a text prompt with a diffusion model. The API returns raw image bytes; this node base64-encodes them at data.image_base64 with data.content_type, ready to pipe into an upload or email node. Image generation runs on GPU and bills by compute time; expect several seconds per image and a large execution record.

Hugging Face: Generate Text

Raw text-generation (completion-style) inference on an hf-inference hosted model: send a prompt, get its continuation at data[0].generated_text. For instruction following prefer the Chat Completion node; this pipeline suits base models like GPT-2. Cold models return 503 with an estimated_time to retry after.

Hugging Face: Like a Repo

Like a model, dataset, or Space as the connected account. Reversible with the Unlike node.

Hugging Face: Summarize Text

Produce a shorter version of a document with a dedicated summarization model. The summary lands at data[0].summary_text. Cold models return 503 with an estimated_time to retry after.

Hugging Face: Transcribe Audio

Transcribe speech to text with a Whisper-family model. Send the audio as a base64-encoded string (pipe it from an upstream file or HTTP node); the transcript lands at data.text. Long recordings take a while - the node allows up to 120 seconds.

Hugging Face: Translate Text

Translate text with a translation model. The result lands at data[0].translation_text. Language pairs are model-dependent: single-pair models (e.g. Helsinki-NLP/opus-mt-en-fr) need no parameters; multilingual models take src_lang and tgt_lang.

Hugging Face: Unlike a Repo

Remove the connected account's like from a model, dataset, or Space. The mirror of the Like node.

Hugging Face: Zero-Shot Classify Text

Classify text against labels you invent on the fly - no training needed. Supply candidate labels like refund or complaint and get a score per label, ideal for routing tickets, tagging form submissions, or triaging content in an automation.

Connecting Hugging Face takes one credential

Hugging Face authenticates everything with a User Access Token created at huggingface.co under Settings and Access Tokens. Tokens start with hf_ and are shown once at creation. A fine-grained token is the right shape for this connector: give it the permission to make calls to Inference Providers plus read access to repos, and add repo write only if you plan to use the like, create repo, or delete repo nodes.

1

At huggingface.co, open Settings, then Access Tokens, create a fine-grained token with the Inference Providers call permission and repo read, and copy it. It is shown once.

2

In Mattering Automations, open Connectors, choose Hugging Face, and create a credential holding the token.

3

Click Test Connection. It calls the free whoami endpoint, so a valid token passes without spending anything.

4

Click Connect. All 36 Hugging Face nodes appear in the canvas palette.

Common uses for the Hugging Face MCP

Chat completions with server-side provider routing

Hugging Face: Chat Completion runs any conversational model served by Inference Providers, and the router picks the fastest live provider by default. Append :cheapest, :preferred, or a provider name like :groq to the model id to steer the routing, and pick models from Hugging Face: List Router Models, which shows each provider's pricing per million tokens, context length, and tool support before you commit.

Embeddings and zero-shot routing without training anything

Hugging Face: Create Embeddings turns one string or an array of them into float vectors for search and RAG, on CPU-friendly sentence-transformers models. Hugging Face: Zero-Shot Classify Text scores text against labels you invent on the fly, such as refund or complaint, which makes it a ready-made router for tickets, form submissions, and inbound content, with Hugging Face: Classify Text covering fixed label sets like sentiment.

An AI research digest that writes itself

Hugging Face: List Daily Papers reads the community-curated Daily Papers feed and Hugging Face: List Trending Repos reads what is rising across models, datasets, and Spaces. Chain either into Hugging Face: Chat Completion or Hugging Face: Summarize Text on a schedule and post the digest to your team every morning.

Images and transcripts as workflow steps

Hugging Face: Generate Image turns a prompt into an image on a diffusion model and hands the result to the next node as base64 with its content type, ready for an upload or email step. Hugging Face: Transcribe Audio runs a Whisper-family model over base64 audio piped from an upstream node and lands the transcript at data.text.

Watching models, datasets, and Spaces like a dashboard

Hugging Face: Search Models filters the catalog by task, author, library, or serving provider, Hugging Face: Get Model returns downloads, likes, license, and gating status for one repo, and Hugging Face: Get Space Runtime reports whether a deployed demo is RUNNING, SLEEPING, or erroring. A scheduled workflow can page you when a Space falls over or when a watched model ships a new revision via Hugging Face: List Model Refs.

Repo housekeeping from a flow

Hugging Face: Create Repo scaffolds a model, dataset, or Space repo, Hugging Face: Delete Repo removes one the token administers, and Hugging Face: Like a Repo and Hugging Face: Unlike a Repo manage the account's likes. File pushes stay in your CI and SDKs by design; the connector handles the lifecycle around them.

Example: The morning AI research digest

Built on the Automations canvas in a few minutes, or described in plain language to Matter and built for you.

1

Trigger: a schedule fires at 8am each weekday.

2

Hugging Face: List Daily Papers pulls the current community-curated feed with titles, summaries, and upvotes.

3

Hugging Face: Chat Completion turns the top entries into a five-bullet briefing, with the feed dropped into the messages array as a template reference.

4

The workflow posts the briefing to your team chat, so the digest exists before anyone has opened a browser.

Hugging Face MCP questions, answered

Hugging Face already has an official MCP server. Why use this one?

It does, at hf.co/mcp, and it is the right choice when an assistant just needs to search the Hub or call Spaces as tools in a conversation. This connector solves a different problem: the same Hugging Face surface as typed nodes on an automation canvas, where scheduled and triggered workflows run with no assistant in the loop, inference results chain into tasks, messages, and database rows, and an MCP-connected assistant can build or edit the whole workflow around them. The two are complementary, and nothing stops you from using both.

Does this work with Claude, ChatGPT, or other AI assistants?

Yes. Mattering exposes these nodes over MCP, an open protocol for connecting AI assistants to tools, so any MCP-capable assistant can connect to your workspace and reach Hugging Face through them. Matter, the assistant built into the platform, uses the same nodes natively. Nothing here is tied to one assistant.

What does it cost to run the inference nodes?

Inference bills to your own Hugging Face account through Inference Providers credits. Hugging Face's pricing docs currently list $0.10 of included monthly credits on free accounts, an amount they mark as subject to change, and $2.00 per month on PRO, with usage past the credits billed pay-as-you-go at the provider's listed price and no markup from Hugging Face. The hf-inference pipelines bill by compute time, and since mid-2025 that provider focuses on CPU tasks like embeddings and classification, so GPU work such as image generation costs more per call. That is all separate from Mattering credits, which are pooled at the organization level and cover running the automation itself.

Are the Hub catalog nodes free to run?

Yes. The 22 query nodes read the Hub API, which is free within rate limits: Hugging Face's docs currently list 1,000 requests per 5 minutes for an authenticated free account, with higher tiers above that. The whoami endpoint behind the credential test counts against the same rate limits, so verify with it rather than polling it.

Which models can the nodes call?

Any model your token can reach, which for public models is effectively the catalog. Model fields are free text with documented defaults, such as openai/gpt-oss-120b on chat and sentence-transformers/all-MiniLM-L6-v2 on embeddings, because the served line-up changes daily. Hugging Face: List Router Models is the discovery surface for chat, and Hugging Face: Search Models with a task and provider filter finds pipeline models that hf-inference actually serves.

Why did a pipeline node return a 503?

The model is cold. hf-inference returns 503 with an estimated load time while it spins a model up, and the connector surfaces that as a retry hint on the execution record so it does not read like an auth failure. Wait roughly the estimated time and run the step again; popular default models are warm far more often than long-tail ones.

What token permissions do the nodes need?

A fine-grained token needs the permission to make calls to Inference Providers for the inference nodes and repo read for the catalog. The like, create repo, and delete repo nodes additionally need write access, and a classic read-only token will return 403 on them while every query node still works. Hugging Face documents token roles on its security tokens page and the Inference Providers permission in its Inference Providers docs.

What is not covered by the 36 nodes?

Deliberate gaps. Uploading files into repos needs a multi-part commit protocol the canvas cannot produce, so content pushes stay in CI and SDKs while repo create and delete are covered. Text to speech and text to video are served only by third-party providers with provider-specific shapes, so they are not shipped. Streaming is out of scope: nodes return one complete response, and the connector strips stream from outbound bodies. Discussions, webhooks management, organization admin, and dedicated Inference Endpoints are real surfaces that sit outside this connector.

36 nodes waiting on the canvas

Put Hugging Face to work inside your workspace

Connect once, then use every node in visual automations and AI agent workflows across the Mattering platform.

All product names, logos, and brands are property of their respective owners and are used for identification purposes only. Use of these names, logos, and brands does not imply endorsement or affiliation. Hugging Face is a product of its respective owner; Mattering is an independent platform.

Start Free Explore our Products