Multi-MCP AI Agent

Multi-MCP AI Agent

An Agentic AI agent system that leverages multiple MCP (Model Control Protocol) servers to provide a wide range of capabilities, from basic mathematical operations to advanced external service integrations like Google Workspace and web scraping. The agent includes a Telegram bot interface and Server-Sent Events (SSE) for real-time communication.

🌟 Features

• Multi-MCP Architecture: Utilizes multiple MCP servers for distributed processing and diverse capabilities
• Cognitive Modules: Implements perception, decision-making, memory, and action modules
• External Service Integration:
  • Google Workspace (Gmail, Google Drive, Google Sheets)
  • Web scraping and content extraction
  • DuckDuckGo search integration
• Real-time Communication:
  • Telegram Bot interface
  • Server-Sent Events (SSE) for live updates
• Core Components:
  • Agent loop management
  • Session handling
  • Context management
  • Strategic decision making

🏗️ Project Structure

├── agent.py              # Main agent entry point
├── core/                 # Core agent components
│   ├── context.py        # Context management
│   ├── loop.py           # Main agent loop
│   ├── session.py        # Session handling
│   └── strategy.py       # Strategic decision making
├── modules/              # Cognitive modules
│   ├── action.py         # Action execution
│   ├── decision.py       # Decision making
│   ├── memory.py         # Memory management
│   ├── model_manager.py  # Model management
│   ├── perception.py     # Input processing
│   └── tools.py          # Tool definitions
├── config/               # Configuration files
│   ├── models.json       # Model configurations
│   └── profiles.yaml     # MCP server profiles
└── mcp_server_*.py       # MCP server implementations

🚀 Getting Started

Prerequisites

• Python 3.11+
• UV package manager
• Telegram Bot Token (for bot functionality)
• Google Cloud credentials (for Google Workspace integration)

Installation

1. Clone the repository:

git clone <repository-url>
cd <repository-name>

2. Create and activate a virtual environment:

uv venv
venv\Scripts\activate  # On Mac: source venv/bin/activate

3. Install dependencies using UV:

uv sync

4. Set up environment variables:
   • Create .env with Gemini API key and Telegram Bot token.
   • Generate credentials.json using Google OAuth client.

Configuration

1. Configure MCP servers in config/profiles.yaml
2. Set up model configurations in config/models.json
3. Configure Google Cloud credentials:
   • Place credentials.json in the root directory
   • Run the application once to generate token.json

🎮 Usage

Starting the Agent

uv run agent.py

Starting the Telegram Bot Server

uv run telegram_sse_server.py

🛠️ MCP Servers

MCP Server 1: Basic Operations

• Mathematical operations
• Image processing
• File operations

MCP Server 2: Document Processing

• Document indexing
• Semantic search
• Content extraction
• Image captioning

MCP Server 3: Web Integration

• DuckDuckGo search
• Web content fetching
• Rate-limited requests

MCP Server 4: Google Workspace

• Gmail integration
• Google Sheets operations
• Google Drive management
• F1 standings fetcher (URL scraper)

📡 Communication

Telegram Bot

• Command handling
• Message processing
• Real-time responses

SSE Server

• Real-time event streaming
• Client connection management
• Event broadcasting

🧠 Cognitive Architecture

The agent implements a cognitive architecture with the following modules:

• Perception: Processes input and extracts relevant information
• Memory: Manages state and historical data
• Decision: Makes strategic decisions based on input and context
• Action: Executes decided actions through appropriate tools

📝 License

MIT License