MCPAuthFuncapp
A secure Model Context Protocol backend service built on Azure Functions that provides authenticated access to Microsoft Graph API. It leverages Azure Active Directory and APIM to enable secure tool discovery and user context management through the OAuth On-Behalf-Of flow.
README
MCPAuthFuncapp
Overview
This Azure Function App implements a secure Model Context Protocol (MCP) backend service with Azure Active Directory authentication. It provides authenticated access to Microsoft Graph API through Azure API Management (APIM) and supports OAuth On-Behalf-Of (OBO) token exchange.
Architecture
[Client App] → [APIM Gateway] → [Azure Functions] → [Microsoft Graph API]
↓ ↓ ↓ ↓
Bearer Token → OBO Exchange → Graph Token → User Data
Features
Authentication Flow
- OAuth 2.0 On-Behalf-Of Flow: APIM policies automatically exchange user bearer tokens for Microsoft Graph API access tokens
- Azure AD Integration: Seamless authentication with Azure Active Directory
- Token Validation: Automatic token validation and refresh handled by APIM policies
Available Endpoints
-
/hello- Simple greeting endpoint- Returns personalized greeting message
- Supports both query parameters and JSON body input
-
/echo- Request echo service- Returns the exact request body sent
- Useful for testing and debugging
-
/tools- MCP tools discovery- Returns available tools and their schemas
- Supports Model Context Protocol standards
-
/me- User profile retrieval- Requires Authentication: Bearer token in Authorization header
- Retrieves authenticated user's profile from Microsoft Graph API
- Automatic token exchange via APIM OBO flow
-
/status- Health check endpoint- Service status and authentication state
- Returns user context information
Security Features
- Token-based Authentication: All sensitive endpoints require valid Bearer tokens
- APIM Policy Protection: OAuth policies protect against unauthorized access
- Pre-authorized Applications: Configured OAuth consent for trusted applications
- Secure Token Exchange: On-Behalf-Of flow maintains security boundaries
APIM Policies
The included APIM policies provide:
- OAuth Authentication: Validates incoming bearer tokens
- Token Exchange: Automatic OBO flow for Microsoft Graph access
- Error Handling: Comprehensive error responses for authentication failures
- Security Headers: Proper CORS and security headers
Configuration
Environment Variables (replace placeholders)
{{CLIENT_ID_PLACEHOLDER}}- Azure AD Application ID{{CLIENT_SECRET_PLACEHOLDER}}- Azure AD Client Secret{{TENANT_ID_PLACEHOLDER}}- Azure AD Tenant ID{{APIM_GATEWAY_URL}}- APIM Gateway URL
Pre-authorized Applications
Configured in preauth.json for OAuth consent bypass (excluded from repository for security).
Development
Prerequisites
- Python 3.9+
- Azure Functions Core Tools
- Azure CLI
- Valid Azure AD App Registration
Local Development
# Install dependencies
pip install -r requirements.txt
# Start local development server
func start
Deployment
- Deploy Azure Function App
- Configure APIM with included policy files
- Update policy placeholders with actual values
- Configure OAuth app registration and permissions
MCP Integration
This service is designed to work as a backend for Model Context Protocol implementations, providing:
- Authenticated Microsoft Graph API access
- Tool discovery and execution
- Secure user context management
Security Notes
⚠️ Important:
- Never commit actual secrets, IDs, or tokens to version control
- Use Azure Key Vault for production secret management
- Regularly rotate client secrets and review OAuth permissions
- Monitor APIM analytics for unusual access patterns
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。
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。
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。
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。
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。
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。
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。
mcp-server-qdrant
这个仓库展示了如何为向量搜索引擎 Qdrant 创建一个 MCP (Managed Control Plane) 服务器的示例。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。