Denser Agent

A multi-agent system built on MCP (Model Context Protocol) architecture providing specialized AI assistants with database analytics, weather information, and meeting scheduling capabilities.

Quick Start

Prerequisites

• Python 3.8+ with pip
• PostgreSQL (for database tools)
• Anthropic API key

pip install -r requirements.txt
pip install -e .

Environment Setup

export CLAUDE_API_KEY="your-anthropic-api-key"
export OPENWEATHER_API_KEY="your-weather-api-key"  # Optional

Database Setup

docker compose up -d postgres
cd tools/database/postgres && python setup_postgres.py

Run Customer Support Agent

1. Start MCP servers (Terminal 1):

   cd agents/customer_support
   python start_servers.py

   This starts all required MCP servers (database, weather, meeting) and keeps them running. You'll see:

   🌟 MCP Servers are running!
   🗄️ Database: http://localhost:8002
   🌤️ Weather: http://localhost:8001
   📅 Meeting: http://localhost:8004
   

   Keep this terminal open - the servers run here.

2. Start the main application (Terminal 2):

   cd /path/to/denser-agent  # Project root
   python -m agents.customer_support.app

   This starts the FastAPI web interface on port 4000.

3. Open your browser: http://localhost:4000

Available Tools

Database Tools (Port 8002)

• PostgreSQL integration with sample data (customers, products, orders)
• SQL query execution with automatic chart generation
• Analytics queries optimized for visualization

AI agent handling a database analytics query with automatic chart generation

cd tools/database/postgres
python postgres_server.py  # Start server
python postgres_demo.py    # Test operations

Weather Tools (Port 8001)

• Current weather data for any location
• Weather forecasts with detailed conditions
• Real or simulated data (with/without API key)

cd tools/weather
python weather_server.py  # Start server
python weather_demo.py    # Test operations

Meeting Tools (Port 8004)

• Meeting scheduling link generation
• Calendar integration via configurable URL
• Simple booking workflow

cd tools/meeting
python meeting_server.py  # Start server
python meeting_demo.py    # Test operations