Malaysia Transit MCP

MCP (Model Context Protocol) server for Malaysia's public transit system, providing real-time bus and train information across 10+ cities in Malaysia.

Data Source: Malaysia Transit Middleware

Table of Contents

• Features

• Architecture

• Quick Start

• Installation

• Configuration

• Available Tools

• Usage Examples

• AI Integration Guide

• Supported Service Areas

• Deployment

• Troubleshooting

• Contributing

Features

• 10 Operational Service Areas across Malaysia

  • Klang Valley (Rapid Rail KL, Rapid Bus KL, MRT Feeder)

  • Penang (Rapid Penang)

  • Kuantan (Rapid Kuantan)

  • Kangar, Alor Setar, Kota Bharu, Kuala Terengganu, Melaka, Johor, Kuching (BAS.MY)

• Real-time Vehicle Tracking - Live positions of buses and trains

• Stop Search & Information - Find stops by name or location

• Route Discovery - Browse available routes with destinations

• Arrival Predictions - Get real-time arrival times at stops

• Multi-Modal Support - Both bus and rail services

• Provider Status Monitoring - Check operational status of transit providers

• Location Detection - Automatically detect service areas using geocoding

Architecture

This MCP server acts as a bridge between AI assistants and the Malaysia Transit Middleware API:

Quick Start

Local Testing with Smithery Playground

Step 1: Start Your Middleware

First, ensure your Malaysia Transit Middleware is running:

The middleware should be running on http://localhost:3000.

Step 2: Configure Environment

Create a .env file in the MCP project root:

Edit .env:

Step 3: Start Smithery Dev Server

This will:

1. Build your MCP server

2. Start the Smithery CLI in development mode

3. Open the Smithery playground in your browser

Step 4: Test in Smithery Playground

In the Smithery playground interface:

1. Test the hello tool:

   Call: hello

   Expected: Returns server info and middleware URL

2. List service areas:

   Call: list_service_areas

   Expected: Returns all available transit areas

3. Search for stops:

   Call: search_stops Parameters: area: "penang" query: "Komtar"

   Expected: Returns matching stops

4. Get real-time arrivals:

   Call: get_stop_arrivals Parameters: area: "penang" stopId: "<stop_id_from_search>"

   Expected: Returns upcoming bus arrivals

Installation

Configuration

Environment Variables

The MCP server uses environment variables for configuration. When deployed to Smithery, set these in the deployment settings:

• MIDDLEWARE_URL (required): Malaysia Transit Middleware API URL

  • Local: http://localhost:3000

  • Production: Your deployed middleware URL (e.g., https://malaysiatransit.techmavie.digital)

• GOOGLE_MAPS_API_KEY (optional): Google Maps API key for location detection

  • If not provided, falls back to Nominatim (free but less accurate)

  • Get your API key from Google Cloud Console

Development

To run the MCP server in development mode:

Build

To build the MCP server for deployment:

Available Tools

Service Area Discovery

list_service_areas

List all available transit areas in Malaysia.

Parameters: None

Returns: List of service areas with their IDs, names, and capabilities.

Example:

get_area_info

Get detailed information about a specific area.

Parameters:

• areaId (string): Service area ID (e.g., "penang", "klang-valley")

Example:

Location Detection

detect_location_area ⭐

Automatically detect which transit service area a location belongs to using geocoding.

Parameters:

• location (string): Location name or place (e.g., "KTM Alor Setar", "Komtar", "KLCC")

Returns: Detected area ID, confidence level, and location details.

Example:

Stop Information

search_stops

Search for stops by name. Use detect_location_area first if unsure about the area.

Parameters:

• area (string): Service area ID

• query (string): Search query (e.g., "Komtar", "KLCC")

Example:

get_stop_details

Get detailed information about a stop.

Parameters:

• area (string): Service area ID

• stopId (string): Stop ID from search results

get_stop_arrivals ⭐

Get real-time arrival predictions at a stop.

Parameters:

• area (string): Service area ID

• stopId (string): Stop ID from search results

Returns: Includes a comprehensive disclaimer about prediction methodology, followed by arrival data with:

• Calculation method (shape-based or straight-line)

• Confidence level (high, medium, or low)

• ETA in minutes

• Vehicle information

Prediction Methodology:

• Shape-Based Distance (Preferred): Uses actual route geometry, accurate within ±2-4 minutes

• Straight-Line Distance (Fallback): Conservative estimates with 1.4x multiplier

• Includes GPS speed validation, time-of-day adjustments, and stop dwell time

• Conservative bias: Better to arrive early than miss the bus

Example:

find_nearby_stops

Find stops near a location.

Parameters:

• area (string): Service area ID

• lat (number): Latitude coordinate

• lon (number): Longitude coordinate

• radius (number, optional): Search radius in meters (default: 500)

Route Information

list_routes

List all routes in an area.

Parameters:

• area (string): Service area ID

get_route_details

Get detailed route information.

Parameters:

• area (string): Service area ID

• routeId (string): Route ID from list_routes

get_route_geometry

Get route path for map visualization.

Parameters:

• area (string): Service area ID

• routeId (string): Route ID from list_routes

Real-time Data

get_live_vehicles ⭐

Get real-time vehicle positions.

Parameters:

• area (string): Service area ID

• type (enum, optional): Filter by type ('bus' or 'rail')

Example:

get_provider_status

Check provider operational status.

Parameters:

• area (string): Service area ID

Testing

hello

Simple test tool to verify server is working.

Usage Examples

Find When Your Bus is Coming

Track Live Buses

Discover Routes

AI Integration Guide

Key Use Cases

1. "When is my bus coming?" ⭐

This is the PRIMARY use case. Users want to know when their next bus/train will arrive.

Workflow:

2. "Where is my bus right now?"

Users want to track their bus in real-time.

Workflow:

Tool Usage Patterns

Always Start with Location Detection

When a user mentions a location without specifying the area, use location detection:

Search Before Details

Always search for stops/routes before requesting details:

Response Formatting

Arrival Times

Format arrival times in a user-friendly way:

Multiple Arrivals

Present multiple arrivals clearly:

Error Handling

Provider Unavailable

Best Practices

1. Use location detection when users mention place names

2. Always specify area for every tool (except list_service_areas and detect_location_area)

3. Search before details - don't guess IDs

4. Handle errors gracefully - providers may have temporary outages

5. Format responses clearly - use minutes, not timestamps

6. Don't cache real-time data - it updates every 30 seconds

Supported Service Areas

Location to Area Mapping

The detect_location_area tool automatically maps common locations to service areas:

Deployment

Deploy to Smithery

This MCP is designed to be deployed to Smithery:

1. Push to GitHub:

   git push origin main

2. Smithery will auto-deploy from your GitHub repository

3. Configure Environment Variables in Smithery:

   • Go to Settings → Environment

   • Add MIDDLEWARE_URL: Your deployed middleware URL

   • Add GOOGLE_MAPS_API_KEY: Your Google Maps API key (optional)

Environment Configuration

Set these environment variables in Smithery deployment settings:

Troubleshooting

Connection Issues

If you can't connect to the middleware:

1. Verify your MIDDLEWARE_URL is correct

2. Ensure the middleware is running and accessible

3. Check network connectivity

4. Test middleware directly: curl https://your-middleware-url/api/areas

No Data Returned

If tools return empty data:

1. Check if the service area is operational using get_provider_status

2. Verify the area ID is correct using list_service_areas

3. Check middleware logs for errors

Real-time Data Unavailable

Real-time data depends on the upstream GTFS providers:

1. Use get_provider_status to check provider health

2. Some providers may have temporary outages

3. Check the middleware logs for API issues

Location Detection Not Working

If location detection returns incorrect results:

1. Ensure GOOGLE_MAPS_API_KEY is set in environment variables

2. Check Google Cloud Console for API quota limits

3. Verify the API key has Geocoding API enabled

4. Falls back to Nominatim if Google Maps fails

Requirements

• Node.js: >= 18.0.0

• Malaysia Transit Middleware: Running instance (local or deployed)

• Google Maps API Key: Optional, for enhanced location detection

Project Structure

Related Projects

• Malaysia Open Data MCP - MCP for Malaysia's open data portal

Contributing

Contributions are welcome! Please feel free to submit pull requests or open issues.

License

MIT - See LICENSE file for details.

Acknowledgments

• Malaysia Open Data Portal for GTFS data

• Prasarana Malaysia for Rapid KL services

• BAS.MY for regional bus services

• Smithery for the MCP framework

• Google Maps Platform for geocoding services

• OpenStreetMap Nominatim for fallback geocoding

Made with ❤️ by Aliff