m365-mcp-server
A production-ready MCP server that provides secure, delegated access to Microsoft 365 services including Email, SharePoint, OneDrive, and Calendar. It enables AI models to search messages, browse files, manage calendar events, and parse document contents using OAuth 2.1 authentication.
README
m365-mcp-server
A production-ready MCP (Model Context Protocol) server for Microsoft 365, providing secure access to Email, SharePoint, and OneDrive through Azure AD/Entra ID authentication with OAuth 2.1 + PKCE.
Features
- Email Access: List folders, search messages, read email content (including shared mailboxes)
- Calendar Access: List calendars, browse events, expand recurring events with date ranges
- SharePoint/OneDrive: Browse sites, drives, folders, and read file content
- Document Parsing: Extracts readable text from PDF, Word, Excel, PowerPoint, CSV, and HTML files
- OAuth 2.1 + PKCE: Secure authentication via Azure AD/Entra ID
- Delegated Permissions: Users access only their authorized content
- Open WebUI Compatible: Works with native MCP or MCPO proxy
- Production Ready: Docker support, security hardening, structured audit logging
- Token Revocation: RFC 7009 compliant token revocation endpoint
Quick Start
1. Azure AD Setup
Follow docs/entra-app-registration.md to create an Azure AD app registration with these permissions:
openid,offline_access(OIDC)User.Read,Mail.Read,Mail.Read.Shared,Files.Read,Sites.Read.All,Calendars.Read(Microsoft Graph)
2. Configuration
Create a .env file:
# Azure AD / Entra ID (required)
AZURE_CLIENT_ID=your-client-id
AZURE_CLIENT_SECRET=your-client-secret
AZURE_TENANT_ID=your-tenant-id
# Server
MCP_SERVER_PORT=3000
MCP_SERVER_BASE_URL=http://localhost:3000
SESSION_SECRET=$(openssl rand -hex 32)
# Optional
LOG_LEVEL=info
REDIS_URL=redis://localhost:6379
# OAuth signing keys (required in production)
# OAUTH_SIGNING_KEY_PRIVATE=<base64-encoded PEM>
# OAUTH_SIGNING_KEY_PUBLIC=<base64-encoded PEM>
3. Run Locally
# Install dependencies
npm install
# Development mode
npm run dev
# Production build
npm run build
npm start
4. Authenticate
- Open
http://localhost:3000/auth/loginin a browser - Sign in with your Microsoft 365 account
- Note the session ID returned after login
Docker Deployment
Basic
cd docker
docker-compose up -d m365-mcp-server redis
With Open WebUI
cd docker
docker-compose --profile with-webui up -d
With MCPO Proxy
cd docker
docker-compose --profile with-mcpo up -d
Open WebUI Integration
Option A: Native MCP (Recommended)
- In Open WebUI, go to Admin Settings > Tools
- Add MCP Server:
{ "url": "http://localhost:3000/mcp", "transport": "streamable-http" } - Complete OAuth login when prompted
Option B: Via MCPO Proxy
- Start MCPO with the provided config:
mcpo --config docker/mcpo-config.json --port 8000 - In Open WebUI, add as OpenAPI Tool:
http://localhost:8000/openapi.json
MCP Tools
Email Tools
| Tool | Description |
|---|---|
mail_list_messages |
List messages with optional filters (supports shared mailboxes) |
mail_get_message |
Get full message details with body (HTML→text), CC/BCC, and attachment metadata |
mail_list_folders |
List mail folders or subfolders (supports shared mailboxes) |
mail_get_attachment |
Read and parse email attachments (PDF, Word, Excel, PowerPoint, CSV, HTML→text). Max 20MB |
All email tools accept an optional mailbox parameter (email address or user ID) to access shared mailboxes. Omit to use your personal mailbox. Requires Mail.Read.Shared permission with admin consent.
SharePoint/OneDrive Tools
| Tool | Description |
|---|---|
sp_list_sites |
Search and list SharePoint sites |
sp_list_drives |
List drives (OneDrive/document libraries) |
sp_list_children |
List folder contents |
sp_get_file |
Get file content with automatic document parsing (PDF, Word, Excel, PowerPoint → text). Max 20MB |
OneDrive Tools
| Tool | Description |
|---|---|
od_my_drive |
Get personal OneDrive info including drive ID and storage quota |
od_list_files |
List files and folders in personal OneDrive (root or subfolder) |
od_get_file |
Get file content by item_id with automatic document parsing (PDF, Word, Excel, PowerPoint). Max 20MB |
od_search |
Search for files in personal OneDrive only |
od_recent |
List recently accessed files |
od_shared_with_me |
List files shared with you by others |
Calendar Tools
| Tool | Description |
|---|---|
cal_list_calendars |
List all calendars with metadata |
cal_list_events |
List events with optional date range (expands recurring events) |
cal_get_event |
Get full event details including body/description |
Requires Calendars.Read permission (no admin consent needed). Provide start_date and end_date to expand recurring events into individual occurrences.
API Endpoints
| Endpoint | Method | Description |
|---|---|---|
/health |
GET | Health check |
/auth/login |
GET | Initiate OAuth login |
/auth/callback |
GET | OAuth callback |
/auth/logout |
GET | Logout and revoke session |
/auth/status |
GET | Check authentication status |
/revoke |
POST | Token revocation (RFC 7009) |
/mcp |
POST | MCP JSON-RPC endpoint |
/mcp |
GET | MCP SSE stream endpoint |
/mcp |
DELETE | Terminate MCP session |
Security
- OAuth 2.1 + PKCE: Required for all authentication flows
- Delegated Permissions Only: No app-only access, read-only Graph scopes (
Mail.Read.Sharedrequires admin consent) - Token Encryption: AES-256-GCM encryption for session tokens at rest
- PII Redaction: Sensitive data (tokens, emails, secrets) filtered from logs
- Structured Audit Logging: Security events logged with correlation IDs
- Rate Limiting: 100 req/min general, 5/hour for client registration
- Security Headers: HSTS, CSP (no unsafe-inline), Permissions-Policy, X-Frame-Options via Helmet
- Input Validation: Zod schemas + regex validation for all Graph API resource IDs
- DCR Protection: Redirect URI pattern whitelist, rate limiting, audit logging
- Production Enforcement: Config validation requires Redis, HTTPS, persistent signing keys
See docs/security/threat-model.md for full security analysis.
Architecture
┌─────────────────────────────────────────────────────────────┐
│ Open WebUI / Client │
└─────────────────────────────┬───────────────────────────────┘
│ MCP Protocol (Streamable HTTP)
▼
┌─────────────────────────────────────────────────────────────┐
│ m365-mcp-server │
│ ┌─────────────┐ ┌─────────────┐ ┌──────────────────────┐ │
│ │ OAuth 2.1 │ │ MCP Handler │ │ Microsoft Graph │ │
│ │ + PKCE │ │ (JSON-RPC) │ │ Client │ │
│ └──────┬──────┘ └─────────────┘ └──────────┬───────────┘ │
└─────────│────────────────────────────────────│──────────────┘
│ │
▼ ▼
┌──────────────────────┐ ┌─────────────────────────┐
│ Azure AD / Entra ID │ │ Microsoft Graph API │
│ (Authorization) │ │ (Data Access) │
└──────────────────────┘ └─────────────────────────┘
Environment Variables
| Variable | Required | Default | Description |
|---|---|---|---|
AZURE_CLIENT_ID |
Yes | - | Azure AD app client ID |
AZURE_CLIENT_SECRET |
Yes | - | Azure AD app client secret |
AZURE_TENANT_ID |
Yes | - | Azure AD tenant ID |
SESSION_SECRET |
Yes | - | Session encryption key (32+ chars) |
MCP_SERVER_PORT |
No | 3000 | Server port |
MCP_SERVER_BASE_URL |
No | http://localhost:3000 | Public URL (HTTPS required in production) |
REDIS_URL |
Prod | - | Redis URL (required in production) |
OAUTH_SIGNING_KEY_PRIVATE |
Prod | - | RSA private key PEM (required in production) |
OAUTH_SIGNING_KEY_PUBLIC |
Prod | - | RSA public key PEM (required in production) |
OAUTH_ALLOWED_REDIRECT_PATTERNS |
No | - | Comma-separated URI patterns for DCR |
LOG_LEVEL |
No | info | Log level (trace/debug/info/warn/error) |
NODE_ENV |
No | development | Environment mode |
FILE_PARSE_TIMEOUT_MS |
No | 30000 | Document parsing timeout |
FILE_PARSE_MAX_OUTPUT_KB |
No | 500 | Max parsed text output size |
Development
# Install dependencies
npm install
# Run tests
npm test
# Run tests with coverage
npm run test:coverage
# Lint
npm run lint
# Type check
npm run typecheck
# Build
npm run build
MCP Registry
This server is published to the MCP Registry. Add to your MCP client:
{
"mcpServers": {
"m365": {
"command": "npx",
"args": ["-y", "@anthropic/m365-mcp-server"],
"env": {
"AZURE_CLIENT_ID": "your-client-id",
"AZURE_CLIENT_SECRET": "your-client-secret",
"AZURE_TENANT_ID": "your-tenant-id",
"SESSION_SECRET": "your-session-secret"
}
}
}
}
Documentation
Supported Document Formats
sp_get_file automatically extracts readable text from these formats:
| Format | Extensions | Library |
|---|---|---|
.pdf |
pdf-parse | |
| Word | .docx, .doc |
mammoth |
| Excel | .xlsx, .xls |
exceljs |
| PowerPoint | .pptx, .ppt |
Built-in ZIP/XML |
| CSV | .csv |
Built-in |
| HTML | .html |
Built-in |
Other binary formats are returned as base64. Parsed text output is limited to 500KB by default.
Known Limitations
- Maximum file download size: 20MB
- Parsed text output capped at 500KB (configurable via
FILE_PARSE_MAX_OUTPUT_KB) - SharePoint site listing requires search query (Graph API limitation)
- Refresh tokens limited to 24 hours for SPA scenarios
- No write operations (read-only by design)
- Access tokens (JWTs) are stateless and cannot be directly revoked (expire naturally)
License
MIT
Contributing
- Fork the repository
- Create a feature branch
- Submit a pull request
Please ensure all tests pass and the code follows the existing style.
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.