Outlook MCP Server
About
Read, search, and manage emails in Microsoft Outlook through a standardized interface.
Details
- Author
- marlonluo2018
- Categories
- Communication, Other, Automation, Search, Knowledge Base
Jump to
Setup
Install Outlook MCP Server in your MCP client (Claude Desktop, Cursor, Windsurf, and others).
Repository: https://github.com/marlonluo2018/outlook-mcp-server
Follow the installation instructions in the repository README, then restart your MCP client.
AI-powered email management for Microsoft Outlook- Search, compose, organize, and batch forward emails with natural language commands.
⭐This saved you time?Star us- takes 2 seconds, helps thousands of Outlook users find this AI email assistant!⭐
🤔 Project Choice: Local vs Cloud Approach
This project uses win32COM which works entirely locally with your Outlook data (more secure), while our newMicrosoft Graph MCP Serverprovides online access to the full Microsoft 365 ecosystem (more features). Choose based on your security and functionality needs.
- ✅ Local Data Processing: All email operations happen locally on your computer - no data leaves your machine
- ✅ Enhanced Security: No cloud authentication or API tokens required - works with your existing Outlook setup
- ✅ Offline Capability: Access and manage emails even without internet connection
- ✅ Privacy-First: Your email data never touches external servers or cloud services
- ✅ Full Microsoft 365 Ecosystem: Access Mail, Calendar, Teams, OneDrive, SharePoint
- ✅ Cross-Platform Support: Works on Windows, macOS, Linux, and mobile devices
- ✅ Modern Authentication: OAuth 2.0 with device code flow
- ✅ Cloud-Scale Performance: No local application dependencies
- ✅ Online Access: Works directly with cloud-based Microsoft 365 services
- Smart Email Search: "Find emails about budget approval from last week"
- AI Email Writing: Draft replies with context-aware suggestions
- Easy Organization: Create folders and move emails with simple commands
- Batch Forwarding: Send emails to 100s of recipients in minutes, not hours
Theagent_prompt_template.mddefines how the AI assistant behaves when managing emails:
- Purpose: Guide the AI's workflow for email search, summarization, and drafting
- Key Rules: AND logic for searches, 5-by-5 email display, confirmation before sending
- Safety: Built-in constraints ensure user control and prevent unauthorized actions
- How to Use: Copy the template content and use it to configure your AI assistant
See thetemplatefor complete behavior guidelines and workflow definitions
Before: "I need to forward this email to 150 team members..."
- Manual: Click Forward → Type each email → Send →2+ hours wasted
- Traditional Outlook: Create distribution list → Add members → Forward →30+ minutes
After: "Forward this email to everyone in team.csv"
- Outlook MCP Server: Load CSV → AI forwards to all →2 minutes total
- Time Saved: 48+ minutes per batch operation!
- ✅ Python 3.8+
- ✅ Microsoft Outlook 2016+ (must be running)
- ✅ Windows 10+
Method 1: UVX (Recommended - Auto Dependencies)
# 1. Install git clone https://github.com/marlonluo2018/outlook-mcp-server.git cd outlook-mcp-server uvx --with "pywin32>=226" --with-editable "." outlook-mcp-server # 2. Configure your AI assistant # Use this in your MCP client settings: { "mcpServers": { "outlook": { "command": "uvx", "args": ["--with", "pywin32>=226", "--with-editable", ".", "outlook-mcp-server"] } } }
# 1. Install git clone https://github.com/marlonluo2018/outlook-mcp-server.git cd outlook-mcp-server pip install -r requirements.txt python -m outlook_mcp_server # 2. Configure your AI assistant # Use this in your MCP client settings: { "mcpServers": { "outlook": { "command": "python", "args": ["-m", "outlook_mcp_server"] } } }
# 1. Install git clone https://github.com/marlonluo2018/outlook-mcp-server.git cd outlook-mcp-server pip install -r requirements.txt # 2. Configure your AI assistant # Use this in your MCP client settings: { "mcpServers": { "outlook": { "command": "python", "args": ["C:\\Project\\outlook-mcp-server\\outlook_mcp_server\\__main__.py"] } } }
Ask your AI assistant: "Show me my recent emails" - if it works, you're ready!
# Test your setup before connecting to AI python -c "import outlook_mcp_server; print('✅ Server module loaded')" # Test Outlook connection (Windows only) python -c "import win32com.client; outlook = win32com.client.Dispatch('Outlook.Application'); print('✅ Outlook connected')"
- Open Claude Desktop settings
- Find "MCP Servers" section
- Click "Add Server" → "Custom"
- Paste the JSON configuration
- Restart Claude Desktop
- Look for "MCP Configuration" or "Server Settings"
- Add the JSON to your client's config file
- Usually located at:~/.config/[client]/mcp.json
Still Stuck?Report an issuewith your error message and setup details.
- Search: Find emails by subject, sender, content, or date range
- Compose: Write new emails with AI assistance
- Reply: Smart replies that understand conversation context
- Batch Forward: Send emails to 100s of recipients from CSV files (saves hours!)
- List: See all your Outlook folders
- Create: Make new folders with simple commands
- Move: Organize emails between folders
- Delete: Remove folders (careful - this is permanent!)
📧 Batch Forwarding: Save Hours on Email Distribution
- Forward weekly reports to your entire team
- Send important updates to all team members
- Distribute project updates to stakeholders
- Send policy updates to all employees
- Forward training materials to departments
- Distribute announcements to company distribution lists
- Load emails: "Show me recent emails" → Emails appear in cache
- Browse results: View 5 emails per page with clear formatting
- Take action: Reply, move, delete, or get AI summary
- Confirm before sending: AI always asks before sending emails
- Understands natural language: "Find urgent emails from my boss"
- Shows email summaries: One-line overview + key action items
- Drafts with context: Replies understand the conversation
- Never sends without permission: Always confirms before sending
email john@company.com jane@company.com team@company.com
- Automatically splits large lists (max 500 per batch)
- Sends via BCC to protect recipient privacy
- Adds your custom message before original email
- Provides delivery confirmation
Manual forwarding: 100 emails × 30 seconds = 50 minutesBatch forwarding: 30 seconds setup + 2 minutes processing = 2.5 minutesYou save: 47.5 minutes per batch!
"Show me emails from last 3 days" "Find emails about project updates" "Draft a reply to John about rescheduling" "Create a folder called 'Work Projects'" "Move email #3 to the Archive folder"
- list_recent_emails_tool(days=7)- Load recent emails (max: 30 days)
- search_email_by_subject_tool("search term")- Search email subjects
- search_email_by_sender_name_tool("sender name")- Search by sender
- search_email_by_body_tool("search term")- Search email content (slower)
- view_email_cache_tool(page=1)- Browse loaded emails (5 per page)
- get_email_by_number_tool(email_number)- Get full email details
- reply_to_email_by_number_tool(email_number, "reply text")- Reply to email
- compose_email_tool("recipient@email.com", "subject", "body")- Send new email
- get_folder_list_tool()-Always use firstto see available folders
- create_folder_tool("folder name")- Create new folder
- move_email_tool(email_number, "target folder")- Move email between folders
- move_folder_tool("source", "target")- Move folders
- Always check folder list firstbefore moving/deleting
- Emails are cached- use numbers from cache for operations
- Never sends without confirmation- AI always asks before sending
- Loads 100 emails in ~2 seconds
- Searches complete in real-time
- All processing happens locally (your data stays private)
- Input Validation: All user inputs are validated before processing
- Custom Error Messages: Clear, actionable error messages for common issues
- Safety Checks: Prevents invalid operations before they cause problems
- 145+ Unit Tests: Every feature thoroughly tested
- Configuration Coverage: All settings and constants validated
- Edge Case Handling: Tested against unusual inputs and scenarios
- Continuous Quality: Tests run on every change to ensure reliability
- Fewer Errors: Validation catches mistakes before they cause problems
- Better Error Messages: Know exactly what went wrong and how to fix it
- Reliable Operation: Comprehensive testing ensures consistent performance
- Safe Operations: Built-in safeguards prevent accidental data loss
- 100% Local Processing: No data leaves your computer
- No Cloud Services: Works entirely offline
- Secure by Design: Uses your existing Outlook installation
- Check theagent prompt templatefor AI assistant setup
- Seeconfiguration examplesfor different installation methods
- Read theWin32COM vs Graph API comparisonfor technical details
- ReviewWin32COM API implementation guidefor performance optimization
- Report issues on theGitHub repository
⭐ Love saving time on email?Star this repo- helps 10,000+ Outlook users discover AI email management! ⭐
💡 Quick star tip: Click the ⭐ button above - it takes 2 seconds and supports open-source email AI!
Local email connectivity for AI agents — read, draft, send, and organize Outlook mail via MCP. Apache-2.0 licensed.
A service for managing multiple email accounts from various providers like 163, Gmail, QQ, and Outlook.
MCP server for Outlook email, calendar, and contacts — let your AI assistant manage your inbox directly from the conversation.
Interact with Microsoft Outlook for calendar management, email operations, and search functionality.
Hosted MCP server that connects your RSS feeds to Claude, ChatGPT, and Gemini — semantic search, story clustering, and scheduled email digests, no local process.
Access your Microsoft 365 mail, calendar, and files using the Microsoft Graph API.
Access and manage your local Microsoft Outlook calendar on Windows.
Connect Claude, ChatGPT, and other AI tools to your Granola meeting notes via MCP. Query your notes, search transcripts, and get meeting insights in your favorite AI assistants.
Connect your Plaud recordings to any MCP-compatible AI client. Search recordings, read transcripts, and generate documents without leaving your AI assistant.
Agent-native forum for the x402/A2A ecosystem. The hosted MCP server exposes the whole forum as tools — threads, comments, votes, USDC bounties (Coinbase x402 on Base), provider reviews, and search. Endpoint: https://api.achivx.com/mcp/ (HTTP, OAuth 2.1).
Sign in to leave a review
Use Google, GitHub, or an email account so ratings stay tied to real people.
No reviews posted yet.


