MCP Integration Servers

This repository contains MCP (Model Control Protocol) servers for integrating various services with Claude Desktop.

Gmail and Google Calendar MCP Server

The Gmail and Google Calendar integration allows Claude to interact with your Gmail account and Google Calendar, enabling email management and calendar operations.

Features

• Gmail Operations:

  • Get latest emails from inbox
  • Search emails using Gmail query syntax
  • Read email content
  • Send emails with CC and BCC support

• Calendar Operations:

  • Search calendar events
  • Create new calendar events
  • View upcoming events

Setup Instructions

1. Prerequisites:

   • Python 3.x
   • MCP package installed
   • Google Cloud Console project with Gmail and Calendar APIs enabled

2. OAuth Credentials:

   • Go to Google Cloud Console
   • Create a new project or select existing one
   • Enable Gmail API and Google Calendar API
   • Create OAuth 2.0 credentials
   • Download credentials and save as credentials.json in the project directory

3. Installation:

   # Install required packages
   pip install google-auth-oauthlib google-auth-httplib2 google-api-python-client
   
   # Install the server in Claude Desktop
   mcp install server.py
   

4. First Run:

   • Run the server: mcp run server.py
   • First run will prompt for OAuth authentication
   • Follow the browser link to authorize the application
   • Token will be saved for future use

Available Tools

Gmail Tools

• get_latest_emails(max_results: int = 10): Get latest emails from inbox
• search_emails(query: str, max_results: int = 10): Search emails using Gmail query
• get_email_content(email_id: str): Get content of a specific email
• send_email(to: str, subject: str, body: str, cc: Optional[str], bcc: Optional[str]): Send email

Calendar Tools

• search_events(query: str, max_results: int = 10, time_min: Optional[str]): Search calendar events
• create_calendar_event(summary: str, start_time: str, end_time: str, description: Optional[str], location: Optional[str]): Create new event

Backend API MCP Server

The Backend API MCP server provides a template for integrating your custom backend API with Claude Desktop.

Features

• User management operations
• Standardized API responses
• Error handling and logging
• Pydantic models for data validation

Setup Instructions

1. Prerequisites:

   • Python 3.x
   • MCP package installed

2. Installation:

   # Install required packages
   pip install pydantic
   
   # Install the server in Claude Desktop
   mcp install backend_server.py
   

3. Configuration:

   • Update the API endpoint configurations
   • Implement actual API calls in tool functions
   • Add authentication if required

Available Tools

• get_users(max_results: int = 10): Get list of users
• create_user(username: str, email: str, password: str): Create new user
• search_users(query: str): Search for users

Development

Running in Debug Mode

# Run with debug logging
export MCP_DEBUG=true
mcp dev server.py

# Run without debug logging
mcp run server.py

Adding New Tools

1. Define Pydantic models for request/response
2. Create new tool function with @mcp.tool() decorator
3. Implement error handling
4. Add logging for debugging

Best Practices

• Use appropriate error handling
• Include comprehensive logging
• Validate input data using Pydantic models
• Follow security best practices for API keys and tokens
• Keep credentials and sensitive data secure

Security Notes

• Store API keys and credentials securely
• Use environment variables for sensitive data
• Never commit credentials to version control
• Implement rate limiting where appropriate
• Follow OAuth best practices

Common Issues

1. Authentication Errors:

   • Check if credentials.json is present
   • Verify OAuth token validity
   • Re-authenticate if token expired

2. API Rate Limits:

   • Implement exponential backoff
   • Handle quota exceeded errors
   • Monitor API usage

3. Connection Issues:

   • Check network connectivity
   • Verify API endpoint availability
   • Confirm firewall settings

Contributing

1. Fork the repository
2. Create feature branch
3. Commit changes
4. Create pull request

License

This project is licensed under the MIT License - see the LICENSE file for details.