Back to projects

Model Context Protocol

MCP servers enabling AI agents to interact with external tools and services.

live infrastructure mcp

Overview

The Model Context Protocol (MCP) is an open protocol that standardizes how applications provide context to Large Language Models (LLMs). This project implements production-ready MCP servers running on Cloud Run, enabling AI agents to securely interact with Google Workspace services and automated survey tools.

MCP allows AI agents to access external tools and services in a standardized, secure way. Rather than building custom integrations for each service, agents can use MCP servers that expose capabilities through a consistent interface.

Architecture

Server Infrastructure:

  • Cloud Run deployment for scalability and reliability
  • Service account management with minimal permission scopes
  • Automatic credential rotation and secure token handling
  • Telemetry logging to BigQuery for monitoring and debugging

Security Model:

  • Each action tagged with user and agent identifiers
  • Scoped permissions per operation (read-only vs. write access)
  • Rate limiting per connector to prevent abuse
  • Error handling and response normalization

Implemented MCP Servers

Google Tools MCP

Unified endpoint providing AI agents access to core Google Workspace services including Gmail, Tasks, Drive, Calendar, and Docs. See the Google Tools MCP project for detailed documentation.

AutoMR MCP

Specialized server for automating the creation and management of Google Forms surveys. See the AutoMR MCP project for detailed documentation.

Technical Implementation

Protocol Features:

  • Standardized request/response formats
  • Session management for stateful interactions
  • Tool discovery and capability negotiation
  • Streaming responses for long-running operations

Deployment Pipeline:

  • Containerized deployment to Cloud Run
  • Environment-based configuration management
  • Automated health checks and monitoring
  • Continuous deployment from version control

Future Development

  • Expanding to additional Google Workspace services (Sheets, Slides)
  • Implementing caching layer for frequently accessed resources
  • Adding support for batch operations
  • Building MCP servers for other service ecosystems (Slack, GitHub, etc.)
  • Developing local development environment for MCP server testing