Auth0 OAuth MCP Server
A Model Context Protocol server that requires user authentication via Auth0 before enabling secure API access on behalf of the authenticated user.
README
Model Context Protocol (MCP) Server
This is a MCP server which will require the user to first authenticate. The MCP server will then be able to call protected APIs on behalf of the user.
Configuration
Todos API
Before you can use the MCP server, you will need to deploy the Todos API as documented here.
Auth0 Configuration
In the Auth0 dashboard, create a new application in the Applications section (type: "Regular Web Application").
<img src="../docs/create-application.jpg" width="500" alt="Create Application">
Once the application is created, configure the following URL as the callback URL when developing locally:
http://localhost:8788/callback
Set up a KV namespace
- Create the KV namespace:
wrangler kv:namespace create "OAUTH_KV" - Update the Wrangler file with the KV ID
Environment Variables
This MCP Server needs a few environment variables to be configured:
| Setting | Description |
|---|---|
AUTH0_DOMAIN |
The domain of your Auth0 tenant (e.g., acme.auth0.com) |
AUTH0_CLIENT_ID |
The Client ID from the Auth0 application you created |
AUTH0_CLIENT_SECRET |
The Client Secret from the Auth0 application you created |
AUTH0_AUDIENCE |
The unique identifier for your API which you registered in the Auth0 dashbaord (e.g., urn:todos-api) |
AUTH0_SCOPE |
The scopes requested by the MCP Server. Here we request basic profile info (openid email profile), refresh token capability (offline_access), and API access (read:todos) |
NODE_ENV |
The environment setting - use development for local development |
API_BASE_URL |
The base URL where your Todos API is running |
Development
Create a .dev.vars file in the root of the project with the following structure:
AUTH0_DOMAIN=yourdomain.us.auth0.com
AUTH0_CLIENT_ID=The Client ID of the application you created in Auth0
AUTH0_CLIENT_SECRET=The Client Secret of the application you created in Auth0
AUTH0_AUDIENCE=urn:todos-api
AUTH0_SCOPE=openid email profile offline_access read:todos
NODE_ENV=development
API_BASE_URL=http://localhost:8789
Testing the MCP Server
To start the MCP server, you can use the following command:
npm run dev
With MCP Inspector you can connect to the MCP server, list the available tools and call them. Make sure to set the transport type to sse and the URL to http://localhost:8788/sse.
<img src="../docs/local.jpg" width="750" alt="MCP Inspector">
Deploying the MCP Server to Cloudflare
To deploy the MCP Server to Cloudflare, you will first need to set the following secrets:
wrangler secret put AUTH0_DOMAIN
wrangler secret put AUTH0_CLIENT_ID
wrangler secret put AUTH0_CLIENT_SECRET
wrangler secret put AUTH0_AUDIENCE
wrangler secret put AUTH0_SCOPE
wrangler secret put API_BASE_URL
Once the secrets are set, you can deploy the API with the following command:
npm run deploy
In the Auth0 dashboard, also make sure to add a new Callback URL for your deployed MCP server, eg:
https://mcp-auth0-oidc.<your-subdomain>.workers.dev/callback
To test this you can now use the Workers AI LLM Playground. Navigate to https://playground.ai.cloudflare.com/ and connect to your MCP server on the bottom left using the following URL pattern:
https://mcp-auth0-oidc.<your-subdomain>.workers.dev/sse
This will open a popup where you can sign in after which you'll be able to use all of the tools.
<img src="../docs/playground.png" width="750" alt="Workers AI LLM Playground">
Troubleshooting
If you encounter any issues while setting up or using the MCP server, here are some troubleshooting steps:
Check Worker Logs
Visit the Cloudflare Workers Logs in your dashboard
Auth0 Dashboard Logs
- Navigate to the Logs section in your Auth0 Dashboard
- Review authentication attempts and failures
<img src="../docs/troubleshooting.png" width="750" alt="Auth0 Logs">
Common Issues
- If authentication fails, verify your Auth0 configuration and secrets
- For connection issues, ensure your Worker is deployed and the domain is correct
- Check that all callback URLs are properly configured in Auth0
- Verify the API_BASE_URL matches your deployed API endpoint
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。