Microsoft Calendar MCP Server
Enables fetching and managing Microsoft Outlook calendar events within date ranges through natural language, using Azure AD authentication and dual transport support for compatibility with Claude Desktop and other MCP clients.
README
Microsoft Calendar MCP Server
A Model Context Protocol (MCP) server for managing Microsoft Outlook calendar events using Microsoft Graph API. This server provides tools for fetching calendar events within specified date ranges and integrates seamlessly with Claude Desktop, MCP Inspector, and other MCP-compatible clients.
🚀 Current Status: Production-ready with dual transport support (stdio and SSE) for maximum compatibility.
Features
- 📅 Fetch calendar events for specific users and date ranges
- 🔐 Secure Microsoft Graph API authentication using Azure AD
- 🌐 Support for timezone-aware operations
- 🔧 FastMCP-based server with dual transport support:
- stdio transport: For Claude Desktop and CLI clients
- SSE transport: For MCP Inspector and web-based clients
- 📊 Detailed event information including attendees and metadata
- 🏥 Health check endpoint for monitoring
Prerequisites
- Python 3.13 or higher
- Azure AD application with appropriate Microsoft Graph API permissions:
Calendars.ReadorCalendars.ReadWriteUser.Read(for accessing user information)
- Service principal credentials (Client ID, Client Secret, Tenant ID)
Installation
Method 1: Using UV (Recommended)
-
Clone the repository:
git clone <repository-url> cd mcp-server-outlook-calendar -
Install dependencies using UV:
uv sync
Method 2: Using pip
-
Clone the repository:
git clone <repository-url> cd mcp-server-outlook-calendar -
Create and activate a virtual environment:
python -m venv .venv source .venv/bin/activate # On Unix/macOS # OR .venv\Scripts\activate # On Windows -
Install dependencies:
pip install -e ".[dev]"
Configuration
-
Create a
.envfile in the project root with your Azure AD credentials:AZURE_TENANT_ID=your_tenant_id_here AZURE_CLIENT_ID=your_client_id_here AZURE_CLIENT_SECRET=your_client_secret_here -
Test your connection:
# Using UV uv run python test_connection.py # Using pip python test_connection.py
Usage
Running the Server
Method 1: SSE Transport (for MCP Inspector)
# Using UV (recommended)
uv run python main.py
# Using pip
python main.py
Server will run on: http://0.0.0.0:8000
SSE Endpoint: http://0.0.0.0:8000/sse
Health Check: http://0.0.0.0:8000/health
Method 2: Stdio Transport (for Claude Desktop)
# Using UV
uv run python -m ms_calendar.server
# Using pip
python -m ms_calendar.server
Running with Claude Desktop (MCP)
To integrate this server with Claude Desktop using MCP (Model Context Protocol):
-
Add the server configuration to your Claude Desktop config file:
On macOS:
~/Library/Application Support/Claude/claude_desktop_config.jsonOn Windows:%APPDATA%\Claude\claude_desktop_config.json{ "mcpServers": { "ms-calendar": { "command": "python", "args": [ "-m", "ms_calendar.server" ], "cwd": "/path/to/your/mcp-server-outlook-calendar" } } }Note: Update the
cwdpath to match your local project location. -
Restart Claude Desktop to load the new server configuration.
The server will be available as a tool within Claude Desktop for calendar operations.
As a Python Library
from ms_calendar.calendar_service import fetch_all_calendar_events, get_graph_client
from datetime import datetime, timedelta
import asyncio
async def main():
# Get a Graph client
graph_client = get_graph_client()
# Fetch events for the next 7 days
start_date = datetime.utcnow()
end_date = start_date + timedelta(days=7)
events = await fetch_all_calendar_events(
graph_client=graph_client,
user_id="user@example.com",
start_date=start_date,
end_date=end_date
)
for event in events:
print(f"Event: {event.subject} - {event.start.date_time} to {event.end.date_time}")
if __name__ == "__main__":
asyncio.run(main())
Running with MCP Inspector
For development and testing, you can use the MCP Inspector with SSE transport:
-
Start the server in SSE mode:
# Using UV (recommended) uv run python main.py # Using pip python main.py -
Run the MCP Inspector:
npx @modelcontextprotocol/inspector -
Configure the inspector:
- Transport Type:
SSE - URL:
http://127.0.0.1:8000/sse - The inspector will automatically open in your browser
- Transport Type:
Note: The SSE transport is currently marked as deprecated in favor of StreamableHttp, but it remains fully functional for MCP Inspector usage.
API Reference
get_calendar_events_time_specific
Fetch calendar events for a user within a specified date range using Microsoft Graph API.
Arguments:
user_id(str): Microsoft Graph user ID or email address (e.g., "user@example.com")start(str): Start date in ISO format (default: "2025-06-16")end(str): End date in ISO format (default: "2025-06-16")timezone(str): IANA timezone identifier (default: "Asia/Manila")
Returns:
dict: JSON object containing:count(int): Number of events foundevents(list): Array of event objects with:id(str): Event unique identifiersubject(str): Event title/subjectattendees(list): Array of attendee objects with email, name, and type
str: Error message if the operation fails
Example Response:
{
"count": 2,
"events": [
{
"id": "AAMkAGI2...",
"subject": "Team Meeting",
"attendees": [
{
"email": "john@example.com",
"name": "John Doe",
"type": "required"
}
]
}
]
}
/health
Health check endpoint for monitoring server status.
HTTP GET /health
Returns JSON:
{"status": "ok"}
Azure AD Setup
To use this server, you'll need to set up an Azure AD application:
-
Register an Application:
- Go to Azure Portal → Azure Active Directory → App registrations
- Click "New registration"
- Provide a name and select account types
- Register the application
-
Configure API Permissions:
- Go to "API permissions" in your app
- Add Microsoft Graph permissions:
Calendars.Read(to read calendar events)User.Read(to read user information)
- Grant admin consent for your organization
-
Create Client Secret:
- Go to "Certificates & secrets"
- Create a new client secret
- Copy the secret value (you won't see it again)
-
Note Application Details:
- Copy the Application (client) ID
- Copy the Directory (tenant) ID
- Use these in your
.envfile
Testing & Coverage
- Tests use
pytest,pytest-asyncio, andpytest-cov - Microsoft Graph API calls can be mocked for testing
- Run all tests with:
pytest - Generate an HTML coverage report with:
pytest --cov=src --cov-report=html open htmlcov/index.html
Transport Support
SSE Transport (Server-Sent Events)
- Entry Point:
main.py - Port: 8000 (default)
- Use Cases: MCP Inspector, web-based clients, development/testing
- Features: Real-time communication, HTTP debugging support
- Health Check:
GET /healthreturns{"status": "ok"}
Stdio Transport
- Entry Point:
ms_calendar.servermodule - Use Cases: Claude Desktop, CLI clients, production deployments
- Features: Lower overhead, direct process communication
Development
This project uses modern Python development tools:
- Package Manager: UV (recommended) or pip
- Framework: FastMCP for MCP server implementation
- API Client: Microsoft Graph SDK for calendar access
- Authentication: Azure Identity for secure Azure AD integration
- Code Quality:
- Black for code formatting (88-char line length)
- Flake8 for linting with custom rules
- MyPy for strict type checking
- Testing: pytest with asyncio support and coverage reporting
- Type Safety: Full mypy support with
py.typedmarker
Project Analysis
For a comprehensive technical analysis of this project, including architecture details, security considerations, and performance characteristics, see ANALYSIS.md.
References
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。