How do AI agents use Hugging Face through Open Connector?

Your user connects Hugging Face once with one of its cataloged authentication methods. Open Connector stores the credential in an encrypted vault and exposes Hugging Face tools to your agent over MCP or a typed API, with credentials injected server-side on each call.

Is this a Hugging Face MCP server?

Yes. Open Connector can serve Hugging Face as a named MCP server with a scoped allowlist and a per-user connection URL, so any MCP client can call Hugging Face actions with credentials injected server-side.

Where do Hugging Face credentials live?

In your own infrastructure. Open Connector keeps credentials in its own vault and injects them at call time, so they never leave your environment.

Hugging Face integration

Hugging Face integration for AI agents.

Hugging Face integration for AI agents with secure authentication and server-side credential injection. Open Connector runs the OAuth, seals the token in an encrypted vault, and serves Hugging Face tools to your agent over MCP or a typed API — credentials injected server-side, every call audited, nothing leaving your infrastructure. Open source (AGPL-3.0) and self-hostable.

View source on GitHub All integrations →

What your agents can do

Real Hugging Face actions, managed and audited.

Your user connects Hugging Face once; your agent can then we’re on a journey to advance and democratize artificial intelligence through open source and open science — scoped to the OAuth permissions you grant and the tool allowlist you configure. Every action is least-privilege and written to a tamper-evident audit trail.

1
Your user grants Hugging Face access once (OAuth) — the token lands in the vault.
2
Your agent calls a tool over MCP or the typed API; Open Connector injects the credential server-side.
3
Every routed call appends a hash-chained audit record — nothing leaves your infra.

Tools & triggers

Supported Hugging Face tools.

314 tools are generated from the published Hugging Face catalog. Descriptions are plain text; each action remains subject to its configured authentication and tool allowlist.

Showing 314 tools. All published catalog entries are included in this page's server-rendered HTML.

Delete bucket: Delete bucket; buckets
Delete a collection: Delete a collection; collections
Delete a collection: Delete a collection; collections
Delete item: Delete an item from a collection; collections
Delete item: Delete an item from a collection; collections
Delete a branch: Delete a branch; datasets
Delete Large file: Delete a Xet/LFS file; datasets
Delete a tag: Delete a tag; datasets
Delete a branch: Delete a branch; models
Delete Large file: Delete a Xet/LFS file; models
Delete a tag: Delete a tag; models
Delete notifications: Delete notifications, either by specifying discussionIds or by applying to all notifications with search parameters; notifications
Delete resource group: Delete a resource group from the organization. Requires the org to have a Team plan or higher.; resource-groups
Remove user: Remove a user from a resource group, revoking the access granted through it. Requires the org to have a Team plan or higher.; resource-groups
Delete a SCIM group: Delete a SCIM group; scim
Delete SCIM user: Removes a user from the organization and deletes any pending invitations for non-managed organizations.; scim
Delete a SCIM group: Delete a SCIM group; scim
Delete a SCIM user: Delete a SCIM user; scim
Delete service account: Delete a service account and all its tokens. This action cannot be undone.; service-accounts
Delete service-account token: Delete an access token from a service account. This action cannot be undone.; service-accounts
Delete a discussion: Delete a discussion; discussions
Delete a discussion: Delete a discussion; discussions
Delete PR ref: Deletes the git ref for a closed/merged pull request to free up storage. LFS files unique to this PR will be garbage collected. The PR page and diff will still be viewable using stored commit data.; discussions
Delete embed: Delete SQL Console embed; sql-console
Delete a scheduled job: Delete a scheduled job; jobs
Delete webhook: Delete webhook; webhooks
Delete a branch: Delete a branch; spaces
Delete Large file: Delete a Xet/LFS file; spaces
Delete secret: Delete Spaces's secret; spaces
Delete a tag: Delete a tag; spaces
Delete variable: Delete Spaces's variable; spaces
Delete Space volumes: Delete Space volumes; spaces
Get agent harnesses: Get the registry of AI agents / harnesses known to the Hub, along with the standard environment variables used to detect them. Clients use this registry to identify which agent they are running in when reporting Hub activity.; agents
Get health: Get health; agentic-provisioning
Get a resource by ID: Get a resource by ID; agentic-provisioning
Get services: Get services; agentic-provisioning
Get avatar: Display the avatar for any user or organization. This endpoint redirects to the avatar URL for either a user or an organization; usersorgs
List namespace buckets: List namespace buckets; buckets
Get bucket details: Get bucket details; buckets
Get resource group: Get resource group; buckets
List files: List files; buckets
Xet read token: Get a read short-lived access token for XET; buckets
Xet write token: Get a write short-lived access token for XET upload; buckets
Get collections: Get collections; collections
Get a collection: Get a collection; collections
Get a collection: Get a collection; collections
Get collection resource group: Get collection resource group; collections
Get collection resource group: Get collection resource group; collections
Get Daily Papers: Get Daily Papers; papers
List commits: List commits; datasets
Get a compare rev: Get a compare rev; datasets
Generate JWT: Generate a JWT token for accessing a repository. Supports optional write access for spaces in dev mode, custom expiration, and encryption.; datasets
Get dataset leaderboard: Returns the evaluation results ranked by score for a dataset/task; datasets
List Large files: List Xet/LFS files for a repo; datasets
Get notebook URL: Get a jupyter notebook URL for the requested file; datasets
List references: List references; datasets
Get resource group: Get resource group; datasets
Get security status: Get the security status of a repo; datasets
List folder content: List the content of a repository tree, with pagination support.; datasets
Get folder size: Get the total size of a repository at a given revision, optionally under a specific subpath. Returns the total size in bytes of all files under the specified path (recursively). If a file is stored via Xet/LFS, the LFS file size is used.; datasets
List access requests: List access requests for a gated repository; datasets
Xet read token: Get a read short-lived access token for XET; datasets
Xet write token: Get a write short-lived access token for XET upload; datasets
Get dataset tags: Get all possible tags used for datasets, grouped by tag type. Optionally restrict to only one tag type; datasets
List docs: Get list of available documentation; docs
Search docs: Search any Hugging Face documentation; docs
Full-text search docs: Full-text search across Hugging Face documentation; docs
Get job hardware: Get available job hardware; jobs
List jobs: List of jobs for an entity; jobs
Count jobs: Count the number of jobs for an entity with optional status stage filter; jobs
Get a job: Get a job; jobs
Stream job events: Stream the events of a job, using SSE; jobs
Stream job logs: Stream the logs of a job, using SSE; jobs
Stream job metrics: Stream the metrics of a job, using SSE; jobs
List kernels: List kernels with optional filtering, sorting, and pagination; kernels
Get kernel: Get kernel; kernels
Get kernel: Get kernel; kernels
List commits: List commits; models
Get a compare rev: Get a compare rev; models
Generate JWT: Generate a JWT token for accessing a repository. Supports optional write access for spaces in dev mode, custom expiration, and encryption.; models
List Large files: List Xet/LFS files for a repo; models
Get notebook URL: Get a jupyter notebook URL for the requested file; models
List references: List references; models
Get resource group: Get resource group; models
Get security status: Get the security status of a repo; models
List folder content: List the content of a repository tree, with pagination support.; models
Get folder size: Get the total size of a repository at a given revision, optionally under a specific subpath. Returns the total size in bytes of all files under the specified path (recursively). If a file is stored via Xet/LFS, the LFS file size is used.; models
List access requests: List access requests for a gated repository; models
Xet read token: Get a read short-lived access token for XET; models
Xet write token: Get a write short-lived access token for XET upload; models
Get model tags: Get all possible tags used for models, grouped by tag type. Optionally restrict to only one tag type; models
List notifications: List notifications for the user; notifications
Export the audit log: Export the audit log events in JSON format for a Team or Enterprise organization. The export is limited to the last 100,000 events.; orgs
Get avatar: Retrieve organization avatar. This endpoint returns a JSON with the avatar URL for the organization. If called with the `Sec-Fetch-Dest: image` header, it instead redirects to the avatar URL; orgs
Get org usage: Get org usage for a given period; orgs
Get session inference usage: Get org inference-provider usage broken down per session id, as a time-series of monthly periods.; orgs
Get resource group usage: Get org usage breakdown per resource group, returned as a time-series of monthly periods. Window is [startDate, endDate], defaults to the current month. Both dates must fall within the last 12 months. Storage values are the peak observed within each monthly period.; orgs
Stream usage: Get live usage for org; orgs
Get org usage: Get org usage for a given period; orgs
Get organization members: Get a list of members for the organization with optional search and pagination.; orgs
Get resource groups: Retrieve accessible resource groups. Get all resource groups the user has access to. Requires the org to have a Team plan or higher.; resource-groups
Get resource group: Retrieve a single resource group by id. Requires the org to have a Team plan or higher.; resource-groups
List SCIM groups: Get a list of SCIM groups. Retrieves a paginated list of all organization groups. If you provide the filter parameter, the resources for all matching groups are returned.; scim
Get a SCIM group: Retrieves a group by its ID. If you provide the `excludedAttributes` parameter, the `members` attribute is not returned.; scim
List SCIM-managed users: Retrieves a paginated list of organization members and pending invitations managed by SCIM for non-managed organizations.; scim
Get a SCIM provisioning user: Retrieves a SCIM user by their ID for non-managed organizations.; scim
List SCIM groups: Get a list of SCIM groups. Retrieves a paginated list of all organization groups. If you provide the filter parameter, the resources for all matching groups are returned.; scim
Get a SCIM group: Retrieves a group by its ID. If you provide the `excludedAttributes` parameter, the `members` attribute is not returned.; scim
Get SCIM Resource Types: Returns the list of SCIM 2.0 resource types supported by this server (User and Group).; scim
Get SCIM Schemas: Returns the SCIM 2.0 schema definitions for User and Group resources.; scim
Get SCIM Schema by ID: Returns a single SCIM 2.0 schema definition by its schema URI.; scim
Get SCIM Configuration: Returns the SCIM 2.0 Service Provider configuration, describing the server's capabilities and supported authentication schemes.; scim
List SCIM users: Retrieves a paginated list of all organization members who have been set up, including disabled users. If you provide the filter parameter, the resources for all matching members are returned.; scim
Get a SCIM user: Retrieves a SCIM user by their ID.; scim
List service accounts: List all service accounts for the organization.; service-accounts
Get service account: Retrieve a single service account and its access token metadata. Token secrets are never returned.; service-accounts
Get network security settings: Get the network security settings for an organization.; orgs
List organization repositories: List organization repositories; orgs
Get social handles: Get an organization's social media handles; orgs
List papers: List arXiv papers sorted by published date; papers
Get a paper: Get a paper; papers
Search papers: Perform a hybrid semantic / full-text-search on papers; papers
Quick search: Quick search for models, datasets, spaces, orgs, users, papers, collections, and buckets; repo-search
Registry token: Mints a short-lived EdDSA JWT for the HuggingFace container registry, verifiable via the JWK at `/api/keys/jwt`.; container
List discussions: Get discussions for a repo; discussions
Get discussion details: Get discussion details; discussions
PR storage estimate: Estimates the LFS storage used by a PR that could be freed if the ref is deleted.; discussions
Resolve a file: This endpoint requires to follow redirection; datasets
Resolve a file: This endpoint requires to follow redirection; models
Resolve a file: This endpoint requires to follow redirection; spaces
List scheduled jobs: List scheduled jobs for an entity; jobs
Get a scheduled job: Get a scheduled job; jobs
Get user usage: Get user usage for a given period; users
Get session inference usage: Get user inference-provider usage broken down per session id; users
Get jobs usage: Get user Jobs usage for current subscription period; users
Stream usage: Get live usage for user; users
Get user usage: Get user usage for a given period; users
Get MCP tools: Get the MCP tools for the current user; users
Stream metrics: Get live usage and running jobs count for the logged-in user, over a single SSE connection; users
List user repositories: List user repositories; users
List webhooks: List webhooks; webhooks
Get webhook: Get webhook; webhooks
List space hardware: Get available space hardware; spaces
List commits: List commits; spaces
Get a compare rev: Get a compare rev; spaces
Stream events: Get status updates for a specific Space in a streaming fashion, with SSE protocol; spaces
Generate JWT: Generate a JWT token for accessing a repository. Supports optional write access for spaces in dev mode, custom expiration, and encryption.; spaces
List Large files: List Xet/LFS files for a repo; spaces
Stream logs: Get logs for a specific Space in a streaming fashion, with SSE protocol; spaces
Stream metrics: Get live metrics for a specific Space in a streaming fashion, with SSE protocol, such as current Zero-GPU usage; spaces
Get notebook URL: Get a jupyter notebook URL for the requested file; spaces
List references: List references; spaces
Get resource group: Get resource group; spaces
Get security status: Get the security status of a repo; spaces
List secrets: List a Space's secret keys. Values are never returned.; spaces
List folder content: List the content of a repository tree, with pagination support.; spaces
Get folder size: Get the total size of a repository at a given revision, optionally under a specific subpath. Returns the total size in bytes of all files under the specified path (recursively). If a file is stored via Xet/LFS, the LFS file size is used.; spaces
List variables: List a Space's variables with their values.; spaces
Xet read token: Get a read short-lived access token for XET; spaces
Xet write token: Get a write short-lived access token for XET upload; spaces
List Space templates: Returns the catalog of official Space templates. The `repoId` of a template can be passed as `template` when creating a Space through the create-repo endpoint.; spaces
Get ZeroGPU quota: Get the authenticated user's current ZeroGPU quota usage and limits; spaces
Get trending: Get the trending repositories; modelsspacesdatasets
Retrieve user avatar: This endpoint returns a JSON with the avatar URL for the user. If called with the `Sec-Fetch-Dest: image` header, it instead redirects to the avatar URL; users
Stream usage: Get live usage for user; users
List user likes: List public repos liked by a user; users
User overview: User overview; users
Get social handles: Get a user's social media handles; users
Get user info: Get information about the user and auth method used; auth
Get file metadata: Returns file metadata including size, hash, and links to XET authentication; buckets
Resolve a file: This endpoint requires to follow redirection; datasets
Export access report: Export a report of all access requests for a gated repository; datasets
Resolve a file: This endpoint requires to follow redirection; models
Export access report: Export a report of all access requests for a gated repository; models
Get user info: Get information about the user. Only available through oauth access tokens. Information varies depending on the scope of the oauth app and what permissions the user granted to the oauth app.; oauth
Resolve a file: This endpoint requires to follow redirection; spaces
Update a collection: Update a collection; collections
Update a collection: Update a collection; collections
Update resource group: Update the name and/or description of a resource group. Requires the org to have a Team plan or higher.; resource-groups
Change user role: Change the role of a user in a resource group. Requires the org to have a Team plan or higher.; resource-groups
Update SCIM group: Update attributes of a SCIM group. Updates individual attributes using Operations format. Just provide the changes you want to make using add, remove (only `members` is supported), or replace operations.; scim
Update SCIM user: Modify individual attributes for non-managed organizations. Only the `active` field can be modified. User profile fields are not editable via SCIM.; scim
Update SCIM group: Update attributes of a SCIM group. Updates individual attributes using Operations format. Just provide the changes you want to make using add, remove (only `members` is supported), or replace operations.; scim
Update SCIM user: Update an attribute of a SCIM user. Modify individual attributes using Operations format. Just provide the changes you want to make using add, remove (only `externalId` is supported), or replace operations. If you set `active` to `false`, the user will be deprovisioned from the organization. Complicated SCIM `path` values are not supported like `emails[type eq 'work'].value`.; scim
Update service-account token: Update the name and/or permissions of an existing service account token.; service-accounts
Update network security: Update the network security settings for an organization.; orgs
Update embed: Update SQL Console embed; sql-console
Update notification settings: Update notification settings for the user; notifications
Update watch settings: Update watch settings for the user. Get notified when discussions happen on your watched items.; notifications
Create account request: Create account request; agentic-provisioning
Create deep link: Create deep link; agentic-provisioning
Provision a resource: Provision a resource; agentic-provisioning
Remove resource: Remove resource; agentic-provisioning
Rotate resource credentials: Rotate resource credentials; agentic-provisioning
Update resource service: Update resource service; agentic-provisioning
Create a new comment: Create a new comment; discussions
Create a new comment: Create a new comment; discussions
Create a new comment: Create a new comment; discussions
Create a new comment: Create a new comment; discussions
Create bucket: Create bucket; buckets
Batch file operations: Accepts NDJSON (newline-delimited JSON) where each line is an addFile, copyFile, or deleteFile instruction. All add/copy operations must come before all delete operations. JSON-lines payload: ```json '{"type":"addFile","path":"...","xetHash":"...","mtime":...,"mtimeNanos":...,"contentType":"..."}' + '{"type":"copyFile","path":"...","xetHash":"...","sourceRepoType":"...","sourceRepoId":"...","mtime":...,"mtimeNanos":...,"contentType":"..."}' + '{"type":"deleteFile","path":"..."}' ```; buckets
Duplicate xet files: Duplicate xet-stored files from this repo (source) into another repo (target) by xet hash, without re-uploading file bytes. The caller must then commit the files with their sha256/size as usual.; buckets
List paths info: List paths info; buckets
Add resource group: Add the repository to a resource group; buckets
Create a collection: Create a collection; collections
Add item: Add an item to a collection; collections
Batch update items: Batch update items in a collection; collections
Set collection resource group: Assign, move, or unassign an organization-owned collection to a resource group. Pass null to unassign. User-owned collections are not eligible.; collections
Add item: Add an item to a collection; collections
Batch update items: Batch update items in a collection; collections
Set collection resource group: Assign, move, or unassign an organization-owned collection to a resource group. Pass null to unassign. User-owned collections are not eligible.; collections
Revoke leaked tokens: Publicly invalidate leaked Hugging Face access tokens. Possession of the raw token value is the only proof required: no authentication is needed, and no rights over the owning account or org are necessary. Each raw token is fully invalidated, the owning user is notified by email. Always returns 202, whether or not any of the provided tokens existed, so the response cannot be used to probe token validity.; tokens
Create branch: Create branch; datasets
Commit: For legacy reason, we support both `application/json` and `application/x-ndjson` but we recommend using `application/x-ndjson` to create a commit. JSON-lines payload: ```json { "key": "header", "value": { "summary": "string (REQUIRED)", "description": "string (OPTIONAL - defaults to empty string)", "parentCommit": "string (OPTIONAL - 40-character hex SHA)" } } { "key": "file", "value": { "path": "string (REQUIRED)", "content": "string (OPTIONAL - required if oldPath not set)", "encoding": "utf-8 | base64 (OPTIONAL - defaults to utf-8)", "oldPath": "string (OPTIONAL - for move/rename operations)" } } { "key": "deletedEntry", "value": { "path": "string (REQUIRED)" } } { "key": "lfsFile", "value": { "path": "string (REQUIRED - max 1000 chars)", "oid": "string (OPTIONAL - required if oldPath not set, 64 hex chars)", "algo": "sha256 (OPTIONAL - only sha256 supported)", "size": "number (OPTIONAL - required if oldPath is set)", "oldPath": "string (OPTIONAL - for move/rename operations)" } } ``` JSON payload: ```json { "summary": "string (REQUIRED)", "description": "string (OPTIONAL - defaults to empty string)", "parentCommit": "string (OPTIONAL - 40-character hex SHA)" "files": [ { "path": "string (REQUIRED)", "content": "string (OPTIONAL - required if oldPath not set)", "encoding": "utf-8 | base64 (OPTIONAL - defaults to utf-8)", "oldPath": "string (OPTIONAL - for move/rename operations)" } ], "deletedEntries": [ { "path": "string (REQUIRED)" } ], "lfsFiles": [ { "path": "string (REQUIRED - max 1000 chars)", "oid": "string (OPTIONAL - required if oldPath not set, 64 hex chars)", "algo": "sha256 (OPTIONAL - only sha256 supported)", "size": "number (OPTIONAL - required if oldPath is set)", "oldPath": "string (OPTIONAL - for move/rename operations)" } ] } ```; datasets
Delete Large files: Delete Xet/LFS files in batch; datasets
Duplicate xet files: Duplicate xet-stored files from this repo (source) into another repo (target) by xet hash, without re-uploading file bytes. The caller must then commit the files with their sha256/size as usual.; datasets
List paths info: List paths info; datasets
Check upload method: Check if a file should be uploaded through the Large File mechanism or directly.; datasets
Add resource group: Add the repository to a resource group; datasets
Squash ref: Squash all commits in the current ref into a single commit with the given message. Action is irreversible.; datasets
Create tag: Create tag; datasets
Batch handle access requests: Accept, reject or reset to pending up to 100 access requests for a single gated repository in one call. The same `status` (and optional `rejectionReason`) is applied to every request in the list.; datasets
Cancel access request: Cancel the current user's access request to a gated repository; datasets
Grant access: Grant access to a user for a gated repository; datasets
Handle access request: Handle a user's access request to a gated repository; datasets
Check access: Check if the user has access to the inference endpoint; inference-endpoints
Check access: Check if the user has access to the inference endpoint; inference-endpoints
Start a job: Start a job; jobs
Check access: Check if the user has access to jobs in the namespace; jobs
Check access: Check if the user has access to jobs in the namespace; jobs
Cancel a job: Cancel a job; jobs
Duplicate a job: Duplicate an existing job, re-using its spec; jobs
Duplicate xet files: Duplicate xet-stored files from this repo (source) into another repo (target) by xet hash, without re-uploading file bytes. The caller must then commit the files with their sha256/size as usual.; kernels
Create branch: Create branch; models
Commit: For legacy reason, we support both `application/json` and `application/x-ndjson` but we recommend using `application/x-ndjson` to create a commit. JSON-lines payload: ```json { "key": "header", "value": { "summary": "string (REQUIRED)", "description": "string (OPTIONAL - defaults to empty string)", "parentCommit": "string (OPTIONAL - 40-character hex SHA)" } } { "key": "file", "value": { "path": "string (REQUIRED)", "content": "string (OPTIONAL - required if oldPath not set)", "encoding": "utf-8 | base64 (OPTIONAL - defaults to utf-8)", "oldPath": "string (OPTIONAL - for move/rename operations)" } } { "key": "deletedEntry", "value": { "path": "string (REQUIRED)" } } { "key": "lfsFile", "value": { "path": "string (REQUIRED - max 1000 chars)", "oid": "string (OPTIONAL - required if oldPath not set, 64 hex chars)", "algo": "sha256 (OPTIONAL - only sha256 supported)", "size": "number (OPTIONAL - required if oldPath is set)", "oldPath": "string (OPTIONAL - for move/rename operations)" } } ``` JSON payload: ```json { "summary": "string (REQUIRED)", "description": "string (OPTIONAL - defaults to empty string)", "parentCommit": "string (OPTIONAL - 40-character hex SHA)" "files": [ { "path": "string (REQUIRED)", "content": "string (OPTIONAL - required if oldPath not set)", "encoding": "utf-8 | base64 (OPTIONAL - defaults to utf-8)", "oldPath": "string (OPTIONAL - for move/rename operations)" } ], "deletedEntries": [ { "path": "string (REQUIRED)" } ], "lfsFiles": [ { "path": "string (REQUIRED - max 1000 chars)", "oid": "string (OPTIONAL - required if oldPath not set, 64 hex chars)", "algo": "sha256 (OPTIONAL - only sha256 supported)", "size": "number (OPTIONAL - required if oldPath is set)", "oldPath": "string (OPTIONAL - for move/rename operations)" } ] } ```; models
Delete Large files: Delete Xet/LFS files in batch; models
Duplicate xet files: Duplicate xet-stored files from this repo (source) into another repo (target) by xet hash, without re-uploading file bytes. The caller must then commit the files with their sha256/size as usual.; models
List paths info: List paths info; models
Check upload method: Check if a file should be uploaded through the Large File mechanism or directly.; models
Add resource group: Add the repository to a resource group; models
Squash ref: Squash all commits in the current ref into a single commit with the given message. Action is irreversible.; models
Create tag: Create tag; models
Batch handle access requests: Accept, reject or reset to pending up to 100 access requests for a single gated repository in one call. The same `status` (and optional `rejectionReason`) is applied to every request in the list.; models
Cancel access request: Cancel the current user's access request to a gated repository; models
Grant access: Grant access to a user for a gated repository; models
Handle access request: Handle a user's access request to a gated repository; models
Change read status: Mark discussions as read or unread. If `applyToAll` is true, all notifications for the user matching the search parameters will be marked as read or unread.; notifications
Create resource group: Create a new resource group in the organization. Requires the org to have a Team plan or higher.; resource-groups
Configure auto-join: Configure the auto-join settings of a resource group. Requires the org to have a Team plan or higher.; resource-groups
Add users: Add one or more organization members to a resource group. Requires the org to have a Team plan or higher.; resource-groups
Create a SCIM group: Creates a new group in the organization. The group name must be unique within the organization.; scim
Create SCIM invitation: Creates an invitation for a user to join the organization. The user must have an existing Hugging Face account.; scim
Create a SCIM group: Creates a new group in the organization. The group name must be unique within the organization.; scim
Create a SCIM user: Creates a new user in the organization. If the user already exists, only `active` field will be updated to provision the user.; scim
Create service account: Create a new service account for the organization.; service-accounts
Create service-account token: Create a new access token for a service account. Store the token securely - it cannot be retrieved later.; service-accounts
Rotate service-account token: Invalidate an existing token and generate a new one with the same permissions. The old token will stop working immediately.; service-accounts
Revoke member token: An org admin can revoke a token's access to the org. The token itself isn't deleted, it still works outside the org. Requires the raw token value. Enterprise only.; orgs
Index a paper: Index a paper from arXiv by its ID. If the paper is already indexed, only its authors can re-index it.; papers
Create a new comment: Create a new comment; discussions
Create a new comment: Create a new comment; discussions
Update paper links: Update the project page, GitHub repository, or submitting organization for a paper. Requires the requester to be the paper author, the Daily Papers submitter, or a papers admin.; papers
Create a new comment: Create a new comment; discussions
Create a new comment: Create a new comment; discussions
Quick search: Quick search for models, datasets, spaces, orgs, users, papers, collections, and buckets; repo-search
Create a new discussion: Create a new discussion; discussions
Create a new comment: Create a new comment; discussions
Merge a pull request: Merge a pull request; discussions
Pin a discussion: Pin a discussion; discussions
Change status: Change the status of a discussion; discussions
Change title: Change the title of a discussion; discussions
Duplicate a repository: Duplicate a repository; repos
Create embed: Create SQL Console embed; sql-console
Create a new repository: Create a new repository; repos
Move repo: Move or rename a repo; repos
Create a scheduled job: Create a scheduled job; jobs
Resume a scheduled job: Resume a scheduled job; jobs
Run job: Trigger a scheduled job run. Trigger a scheduled job to run immediately. Throws an error if an instance is already running and job spec does not allow concurrent runs.; jobs
Update job schedule: Update job schedule; jobs
Suspend a scheduled job: Suspend a scheduled job; jobs
Claim paper authorship: Claim paper authorship; papers
Create webhook: Create webhook; webhooks
Update webhook: Update webhook; webhooks
Enable/disable webhook: Enable/disable webhook; webhooks
Replay webhook log: Replay webhook log; webhooks
Create branch: Create branch; spaces
Commit: For legacy reason, we support both `application/json` and `application/x-ndjson` but we recommend using `application/x-ndjson` to create a commit. JSON-lines payload: ```json { "key": "header", "value": { "summary": "string (REQUIRED)", "description": "string (OPTIONAL - defaults to empty string)", "parentCommit": "string (OPTIONAL - 40-character hex SHA)" } } { "key": "file", "value": { "path": "string (REQUIRED)", "content": "string (OPTIONAL - required if oldPath not set)", "encoding": "utf-8 | base64 (OPTIONAL - defaults to utf-8)", "oldPath": "string (OPTIONAL - for move/rename operations)" } } { "key": "deletedEntry", "value": { "path": "string (REQUIRED)" } } { "key": "lfsFile", "value": { "path": "string (REQUIRED - max 1000 chars)", "oid": "string (OPTIONAL - required if oldPath not set, 64 hex chars)", "algo": "sha256 (OPTIONAL - only sha256 supported)", "size": "number (OPTIONAL - required if oldPath is set)", "oldPath": "string (OPTIONAL - for move/rename operations)" } } ``` JSON payload: ```json { "summary": "string (REQUIRED)", "description": "string (OPTIONAL - defaults to empty string)", "parentCommit": "string (OPTIONAL - 40-character hex SHA)" "files": [ { "path": "string (REQUIRED)", "content": "string (OPTIONAL - required if oldPath not set)", "encoding": "utf-8 | base64 (OPTIONAL - defaults to utf-8)", "oldPath": "string (OPTIONAL - for move/rename operations)" } ], "deletedEntries": [ { "path": "string (REQUIRED)" } ], "lfsFiles": [ { "path": "string (REQUIRED - max 1000 chars)", "oid": "string (OPTIONAL - required if oldPath not set, 64 hex chars)", "algo": "sha256 (OPTIONAL - only sha256 supported)", "size": "number (OPTIONAL - required if oldPath is set)", "oldPath": "string (OPTIONAL - for move/rename operations)" } ] } ```; spaces
Delete Large files: Delete Xet/LFS files in batch; spaces
Duplicate xet files: Duplicate xet-stored files from this repo (source) into another repo (target) by xet hash, without re-uploading file bytes. The caller must then commit the files with their sha256/size as usual.; spaces
List paths info: List paths info; spaces
Check upload method: Check if a file should be uploaded through the Large File mechanism or directly.; spaces
Add resource group: Add the repository to a resource group; spaces
Upsert secret: Upsert Spaces's secret; spaces
Squash ref: Squash all commits in the current ref into a single commit with the given message. Action is irreversible.; spaces
Create tag: Create tag; spaces
Upsert variable: Upsert Spaces's variable; spaces
Request access: Request access to a gated repository. The fields requested by repository card metadata (https://huggingface.co/docs/hub/en/models-gated#customize-requested-information); datasets
Request access: Request access to a gated repository. The fields requested by repository card metadata (https://huggingface.co/docs/hub/en/models-gated#customize-requested-information); models
Initiate device authorization: Initiate device authorization; oauth
Register a new OAuth app: Register a new OAuth app; oauth
Get user info: Get information about the user. Only available through oauth access tokens. Information varies depending on the scope of the oauth app and what permissions the user granted to the oauth app.; oauth
Update bucket settings: Update bucket settings; buckets
Update repo settings: Update the settings of a repo; datasets
Update job exposed ports: Replace the exposed ports of a running job, applied live (no re-run). Ports listed in `portsPublic` are reachable without any authentication.; jobs
Update job labels: Replace user-provided labels on a job; jobs
Update repo settings: Update the settings of a repo; models
Change member role: Change the role of a member in the organization. Need a paid plan.; orgs
Update a SCIM group: Updates a group by its ID. The group name must be unique within the organization.; scim
Update a SCIM user: Updates a provisioned user's invitation for non-managed organizations. User profile fields are not editable via SCIM for non-managed organizations.; scim
Update a SCIM group: Updates a group by its ID. The group name must be unique within the organization.; scim
Update a SCIM user: Updates a provisioned user, you'll need to provide all their information fresh - just like setting them up for the first time. Any details you don't include will be automatically removed, so make sure to include everything they need to keep their account running smoothly. Setting `active` to `false` will deprovision the user from the organization.; scim
Update scheduled job labels: Replace user-provided labels on a scheduled job; jobs
Update repo settings: Update the settings of a repo; spaces
Set Space volumes: Set Space volumes; spaces

FAQ

Hugging Face integration, answered

How do AI agents use Hugging Face through Open Connector?: Your user connects Hugging Face once with one of its cataloged authentication methods. Open Connector stores the credential in an encrypted vault and exposes Hugging Face tools to your agent over MCP or a typed API, with credentials injected server-side on each call.
Is this a Hugging Face MCP server?: Yes. Open Connector can serve Hugging Face as a named MCP server with a scoped allowlist and a per-user connection URL, so any MCP client can call Hugging Face actions with credentials injected server-side.
Where do Hugging Face credentials live?: In your own infrastructure. Open Connector keeps credentials in its own vault and injects them at call time, so they never leave your environment.

Give your agents Hugging Face — keep the keys.

Open source, self-hostable, with Hugging Face credentials that never leave your infrastructure. Run it from source today.

View source on GitHub How the MCP gateway works →