mithileshchellappan/traffic-boy-poke

Traffic Boy - Poke MCP Server

Building my First Ever MCP Server in under 3 hours for the Poke MCP Challenge.

A Model Context Protocol (MCP) server that provides live and forecast traffic data using Google Maps API.

✨ Features

• Live Traffic Data: Get real-time traffic conditions between any two locations
• Traffic Forecasting: Predict future travel times with traffic considerations
• Traffic Comparison: Compare current vs. forecasted traffic for route planning
• Flexible Location Input: Supports both addresses and latitude/longitude coordinates
• Multiple Travel Modes: Driving, walking, bicycling, and transit options

🚀 Planned Features

• Search places so you can just say "Apple Store, Bengaluru" and it'll get the correct location
• Access Saved places on Google Maps. Requires a setup to get and store user's oauth tokens.
• Access Live Location.

🛠️ Setup Instructions

1. Prerequisites

• Node.js 18+ installed
• Google Maps API key with the following APIs enabled:
  • Maps JavaScript API
  • Directions API
  • Geocoding API
  • Distance Matrix API

2. Get Google Maps API Key

1. Go to Google Cloud Console
2. Create a new project or select existing one
3. Enable the required APIs mentioned above
4. Create credentials (API Key)
5. Optionally restrict the API key to your domain for security

3. Install Dependencies

npm install

4. Set Environment Variable

export GOOGLE_MAPS_API_KEY="your-api-key-here"

Or create a .env file:

GOOGLE_MAPS_API_KEY=your-api-key-here

5. Run the Server

npm start

🎯 Available Tools

get_live_traffic

Get current traffic conditions between two locations.

Parameters:

• origin: Origin location (address like "New York, NY" or coordinates "40.7128,-74.0060")
• destination: Destination location (same format as origin)
• mode: Travel mode - "driving", "walking", "bicycling", or "transit" (default: "driving")

get_forecast_traffic

Get forecasted traffic conditions for a specific departure time.

Parameters:

• origin: Origin location
• destination: Destination location
• departure_time: ISO timestamp (e.g., "2024-01-15T09:00:00Z") or "now"
• mode: Travel mode (default: "driving")

get_traffic_comparison

Compare current traffic vs. forecasted traffic for decision making.

Parameters:

• origin: Origin location
• destination: Destination location
• forecast_hours: Hours ahead to check (1-24, default: 1)
• mode: Travel mode (default: "driving")

📱 Integration with Poke

1. Add to Poke

1. Go to Poke Settings > Connections > Integrations
2. Add your MCP server endpoint
3. Use the provided configuration in poke-config.json

2. Create Automations

Here are some example automations you can create:

Daily Commute Check

• Trigger: Schedule (8:00 AM weekdays)
• Action: Get live traffic from home to office
• Notification: Send email/SMS with travel time

Smart Departure Time

• Trigger: Calendar event starting soon
• Action: Check traffic and suggest optimal departure time
• Smart Logic: If traffic is bad, send reminder 15 minutes early

Traffic Alert System

• Trigger: Significant traffic delay detected
• Action: Notify with alternative routes
• Integration: Connect with calendar for rescheduling

📝 Example Usage with Poke 🌴

Ask Poke to remember your favourite places like:

• "Remember 323 Merlin Dr California as my Home"
• "Remember Apple Park, Cupertino as my Work"

Ask Poke to remember your routine and set automation to use Traffic Boy like:

• "I leave to work on weekdays at 9 AM, be sure to check for traffic at that time"
• "I go to the gym at 6 PM from work at weekdays, and at 5 PM on weekends directly from home, be sure to check for traffic at that time"

📝 Example Usage

// Get live traffic
{
  "tool": "get_live_traffic",
  "origin": "Times Square, New York, NY",
  "destination": "JFK Airport, New York, NY",
  "mode": "driving"
}

// Get forecast for tomorrow morning
{
  "tool": "get_forecast_traffic",
  "origin": "40.7589,-73.9851",
  "destination": "40.6413,-73.7781",
  "departure_time": "2024-01-15T08:00:00Z"
}

// Compare current vs future traffic
{
  "tool": "get_traffic_comparison",
  "origin": "Boston, MA",
  "destination": "Providence, RI",
  "forecast_hours": 2
}

🔧 Configuration

The server uses the following configuration:

• Port: Runs on stdio (standard MCP protocol)
• API Key: Required via GOOGLE_MAPS_API_KEY environment variable
• Traffic Model: Uses "best_guess" for current, "pessimistic" for forecasts
• Supported Modes: driving, walking, bicycling, transit

🚀 Deployment

For production deployment:

1. Set up your Google Maps API key securely
2. Deploy to a server with Node.js support
3. Configure Poke to connect to your MCP server endpoint
4. Monitor API usage and costs

📊 API Limits & Costs

• Google Maps has generous free tier (up to $200/month credit)
• Directions API: 40,000 requests/month free
• Monitor usage in Google Cloud Console
• Costs scale with usage beyond free tier

🤝 Contributing

Feel free to enhance this MCP server with:

• Additional traffic data sources
• Route optimization algorithms
• Integration with other mapping services
• Advanced analytics and predictions

📄 License

MIT License