Auth0 MCP Server
Official Auth0 MCP server: tenant applications, Actions, logs and users via the Management API
Install
npx @auth0/auth0-mcp-server init --client claude-codeOffers to write the server to ~/.claude.json or the project's .mcp.json.
This is third-party code. Review the repository files before installing.
What it does
The server gives an agent access to an Auth0 tenant's Management API. The agent can create applications and get their domain and client ID, write and deploy Actions, and search logs, for example logins from a specific IP. Login runs in the browser via the OAuth device flow, and the init command writes the server into the chosen client's config. The --read-only and --tools flags limit the available tools.
Who it is for. For developers and admins who configure authentication in Auth0.
Good fit when
- You need to create an Auth0 application for a new project quickly
- You need to find the cause of failed logins in logs
- You need to write and deploy an Action
Not a fit when
- A production tenant where changes need review
- Your authentication is not built on Auth0
Example request
Check Auth0 logs for logins from IP 192.108.92.3 in the last dayLimitations
The project is in beta and its authors do not recommend it for production workloads. Requires Node.js 18 or newer and an Auth0 account with the right permissions. Access from Russia is not guaranteed.
How to disable. Remove the auth0 block from the client config. Revoke the granted access in the Auth0 dashboard.
MCP
- Transport
- stdio
- Authentication
- OAuth
| Environment variables | |
|---|---|
| DEBUG | Enables debug output, auth0-mcp in examples |
Security check
- Changes authentication tenant settings and deploys Action code
- Has access to users and login logs
README in short
The README warns about the beta status and shows installation with the init command for Claude Desktop, Claude Code, Windsurf, Cursor, VS Code and Gemini CLI, plus manual setup for Codex and other clients. It covers login via the OAuth device flow and tool filtering with patterns. There are security recommendations for tool access. MIT licensed.
FAQ
How do I limit the agent to reading?
Add --read-only or the --tools 'auth0_list_*,auth0_get_*' pattern when running init.
Related
NVIDIA's open stack for running OpenClaw, Hermes and LangChain Deep Agents in OpenShell sandboxes with network policy and managed inference
Static code analysis with rules that look like source code, plus a built-in MCP server for AI agents
Trail of Bits skills
Trail of Bits Skills Marketplace
Trail of Bits plugin marketplace for security work: smart contracts, C/C++ and Rust review, Semgrep, CodeQL and fuzzing
Nine skills that let an agent run Strix pentests, fix validated findings and add security scanning to CI