Hugging Face MCP integration for AI agents.
Connect AI agents to Hugging Face through 14 structured actions, including get current user, generate chat completion, and generate embeddings. Review authen…
- ACTION CONTRACTS
- 14
- AUTHENTICATION
- OAuth 2.0 + API key
- SOURCE STATUS
- Provider-backed
CATALOG SOURCE REVIEWED AUGUST 23, 2026 / ACTION NAMES AND SCHEMAS DERIVED FROM WORKING MACHINES PROVIDER SOURCE
CAPABILITY PROFILE
What the Hugging Face integration exposes
The Hugging Face Agent App exposes 14 provider-backed actions for ai and developer tools work. Its current contract lets an authorized agent get the current authenticated hugging face user profile, generate a chat completion with hugging face inference providers, and generate text embeddings with hugging face inference. These operations are called through Working Machines as typed capabilities rather than through browser navigation or copied UI steps.
A connection uses OAuth 2.0 and API key. Before execution, the agent can inspect the selected action, its required fields, declared scopes, and expected output contract. Provider credentials remain inside the Working Machines runtime; the calling agent receives the capability and its structured result, not the underlying secret.
For reliable operation, start with the narrowest action that satisfies the task, resolve stable provider identifiers before changing state, and validate the returned object or status after execution. Availability still depends on the connected Hugging Face account, granted provider permissions, workspace policy, region, plan, and upstream API behavior.
VERIFIED ACTION SAMPLE
Real Hugging Face capabilities.
Showing 12 of 14 actions. Risk labels are conservative signals based on operation names, not substitutes for provider documentation or runtime policy.
get_current_userREADGet the current authenticated Hugging Face user profile.
generate_chat_completionREVIEWGenerate a chat completion with Hugging Face Inference Providers.
INPUTS: model / messages / max_tokens / temperature / top_p
generate_embeddingsREVIEWGenerate text embeddings with Hugging Face inference.
INPUTS: model / inputs
get_dataset_first_rowsREADPreview the first rows of a dataset split from the Hugging Face Dataset Viewer.
INPUTS: dataset / config / split
get_dataset_infoREADGet detailed metadata for a Hugging Face dataset by dataset id.
INPUTS: dataset / config
get_dataset_statisticsREADGet column statistics for a dataset split from the Hugging Face Dataset Viewer.
INPUTS: dataset / config / split
get_model_infoREADGet detailed metadata for a Hugging Face model by modelId.
INPUTS: modelId
get_space_infoREADGet detailed metadata for a Hugging Face Space by repo id.
INPUTS: repoId / revision
get_trendingREADGet trending Hugging Face repositories across models, datasets, and Spaces.
INPUTS: type / limit
list_datasetsREADList Hugging Face datasets using user-friendly search filters.
INPUTS: search / author / filter / sort / direction
list_endpointsREADList Hugging Face Inference Endpoints for a namespace.
INPUTS: namespace / search / tags / limit / cursor
list_modelsREADList Hugging Face models using user-friendly search filters.
INPUTS: search / author / task / limit
INPUT CONTRACTS
Know what the action needs before it runs.
generate_chat_completion
modelREQUIRED- A non-empty string.
messagesREQUIRED- The ordered conversation messages.
max_tokensOPTIONAL- The maximum number of output tokens to generate.
temperatureOPTIONAL- The sampling temperature.
top_pOPTIONAL- The nucleus sampling threshold.
generate_embeddings
modelOPTIONAL- The embedding model identifier.
inputsREQUIRED- The list of texts to embed.
get_dataset_first_rows
datasetREQUIRED- The full dataset identifier, such as owner/dataset.
configREQUIRED- The dataset config or subset name.
splitREQUIRED- The dataset split name to preview.
get_dataset_info
datasetREQUIRED- The full dataset identifier, such as owner/dataset.
configOPTIONAL- The optional dataset config or subset name.
PROVIDER-SPECIFIC WORKFLOWS
Jobs this Agent App can support
Inspect Get Current User
Get the current authenticated Hugging Face user profile. Use this as a bounded discovery step, retain the returned identifier, and avoid expanding the read beyond the task's stated scope.
get_current_userVerify with Get Dataset First Rows
Preview the first rows of a dataset split from the Hugging Face Dataset Viewer. Compare the returned provider state with the intended outcome and preserve stable IDs or canonical links in the run record.
get_dataset_first_rowsSAFETY BOUNDARY
Operate Hugging Face with explicit limits
- Authorize Hugging Face with OAuth 2.0 and API key and grant only the provider access required by the selected actions.
- The current action names appear read-oriented, but returned Hugging Face data may still be sensitive and should be minimized before it enters model context.
- Do not infer permission from catalog visibility. Workspace policy, connection identity, and upstream authorization still govern execution.
- The sampled contract declares provider scopes including `inference-api`, `openid`, `profile`, and `read-repos`; verify the complete action contract before execution.
- After a call, inspect the structured result and execution record before reporting that the Hugging Face task completed successfully.
CONNECTION MODEL
OAuth 2.0 + API key
Hugging Face user access token sent as a Bearer token. Create one from your Hugging Face access token settings.
Working Machines stores provider credentials behind the execution boundary. An agent can use an authorized connection identity, but catalog discovery alone does not reveal OAuth tokens, API keys, or provider secrets.
- READ SIGNALS
- 12
- WRITE SIGNALS
- 0
- HIGH IMPACT
- 0
- REVIEW SIGNALS
- 2
EVIDENCE AND AVAILABILITY
Provider reference
Action names, input fields, authentication types, and counts on this page are generated from the Working Machines provider catalog. Provider behavior, quotas, object semantics, account eligibility, and regional availability remain governed by Hugging Face.