Hugging Face's official remote MCP server — search and inspect models, datasets and Spaces, browse the Hub filesystem over hf:// URIs, and turn Gradio Spaces into callable tools.
Summary
WorkOS's official remote MCP server — run your WorkOS dashboard from an agent: organizations, users, SSO connections, Directory Sync, AuthKit branding and audit log streams.
Features
- Manage organizations, users, SSO connections and Directory Sync
- Edit AuthKit branding and configure audit log streams
- Agent inherits your exact dashboard role — never more access than you have
- Environment-scoped and sandbox by default; production only on request
- Destructive operations require explicit confirmation after a described impact
- API keys and client secrets stripped from responses before reaching the model
- Team-level admin switches: enable, allow production, allow write
- OAuth over streamable HTTP; 15+ documented clients
Installation
Set up this MCP server in your favorite AI agent — copy a ready-made configuration below.
Description
WorkOS MCP is a remote management server at https://mcp.workos.com/mcp that hands an AI agent the same WorkOS dashboard you use — organizations, users, SSO connections, Directory Sync, AuthKit branding, audit logs — and lets it call the WorkOS API on your behalf. Ask it to onboard a customer from a CSV, work out why org_123 cannot complete an SSO sign-in, or set up a Datadog audit log stream, and it works out which operations to run.
The permission model is the reason to look at it. The agent authenticates as you, over OAuth via WorkOS Connect, and inherits your exact dashboard role — a member without admin rights cannot use it to do admin-only things, and a read-only role cannot change anything. It is scoped to one environment at a time and defaults to sandbox, only touching production when you direct it there. Irreversible operations (deleting an organization, connection or directory) are never executed on the first attempt: the agent gets a description of exactly what would be destroyed and has to confirm. API keys, client secrets and signing certificates are stripped from responses before they reach the model's context, so a secret cannot leak into a chat transcript — though the agent can still set a value you supply.
Admin controls. Team admins get three independent switches in team authentication settings, all on by default: enable MCP at all, allow production access, and allow write access. Turning off write access makes the whole surface read-only for everyone.
Deliberate gaps. No user impersonation is exposed and mutations are blocked on impersonated sessions; the agent cannot change the MCP access settings that govern itself, mint or rotate credentials, or delete your WorkOS team; and it is scoped to the single team you authenticate with.
Clients. Streamable HTTP with OAuth, documented for Claude Code, Claude Desktop, ChatGPT (Pro, Business and Enterprise custom connectors), Cursor, VS Code, Zed, Windsurf, Codex CLI, Factory, Goose, OpenCode and Antigravity — 15+ in total, plus any other client that accepts a remote MCP URL.
Related MCP Servers
Resend's official MCP server for transactional and marketing email — send, schedule and batch messages, manage templates, contacts, broadcasts, automations, domains and suppressions from an agent.
Postman's official MCP server — reach your workspaces, collections, specs and environments from Claude Code, Cursor, VS Code or Copilot, in four tool tiers.
Drive live Jupyter notebooks from an agent — read and edit cells, execute code, see rich outputs, and run against local JupyterLab or a cloud sandbox.