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.

MCP Endpoint: https://mcp.techmavie.digital/malaysiatransit/mcp

Analytics Dashboard: https://mcp.techmavie.digital/malaysiatransit/analytics/dashboard

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

• 11 Operational Service Areas + 1 Under Maintenance + 1 Coming Soon across Malaysia
  • Klang Valley (Rapid Rail KL, Rapid Bus KL, MRT Feeder)
  • Penang (Rapid Penang, Penang Ferry, KTM Komuter Utara)
  • Ipoh (BAS.MY Ipoh, KTM Komuter Utara)
  • Seremban (BAS.MY Seremban, KTM Intercity)
  • Kangar (BAS.MY Kangar, KTM Komuter Utara)
  • Alor Setar (BAS.MY Alor Setar, KTM Komuter Utara)
  • Kota Bharu (BAS.MY Kota Bharu, KTM Intercity)
  • Kuala Terengganu, Melaka, Johor, Kuching (BAS.MY only)
  • Kuantan (Under Maintenance)
  • Kota Kinabalu (Coming Soon)
• 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 (shape-based, 40-60% more accurate)
• Schedule Information - Get departure times, route schedules, and operating status
• Fare Calculator - Calculate bus fares for BAS.MY and Rapid Penang routes
• Multi-Modal Support - Bus, rail, and ferry services
• Provider Status Monitoring - Check operational status of transit providers
• Location Detection - Automatically detect service areas using geocoding
• 🆕 KTM Komuter Utara - 23 stations from Padang Besar to Ipoh with fare calculation
• 🆕 KTM Intercity - SH and ERT routes (Tumpat - Gemas - JB Sentral)
• 🆕 Penang Ferry - Butterworth to George Town ferry information
• 🆕 API Analytics - View API usage statistics, endpoint metrics, and client tracking
• 🆕 Analytics Dashboard - Visual dashboard with charts for MCP server usage monitoring
• 🆕 Client Identification - MCP identifies itself to middleware for analytics tracking
• 🆕 Auto-Deployment - GitHub Actions workflow for automatic VPS deployment

Architecture

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

AI Assistant (Claude, GPT, etc.)
    ↓
Malaysia Transit MCP Server (identifies as "Malaysia-Transit-MCP")
    ↓
Malaysia Transit Middleware API (tracks usage via X-App-Name header)
    ↓
Malaysia Open Data Portal (GTFS Static & Realtime)

Client Identification: This MCP automatically sends an X-App-Name: Malaysia-Transit-MCP header with every API request, allowing the middleware to track usage from this MCP separately in its analytics dashboard.

Quick Start

Connect to the MCP Server

Add this configuration to your MCP client (Claude Desktop, Cursor, Windsurf, etc.):

{
  "mcpServers": {
    "malaysia-transit": {
      "transport": "streamable-http",
      "url": "https://mcp.techmavie.digital/malaysiatransit/mcp"
    }
  }
}

Test with MCP Inspector

npx @modelcontextprotocol/inspector
# Select "Streamable HTTP"
# Enter URL: https://mcp.techmavie.digital/malaysiatransit/mcp

Test with curl

# List all available tools
curl -X POST https://mcp.techmavie.digital/malaysiatransit/mcp \
  -H "Content-Type: application/json" \
  -d '{"jsonrpc":"2.0","id":1,"method":"tools/list"}'

# Call the hello tool
curl -X POST https://mcp.techmavie.digital/malaysiatransit/mcp \
  -H "Content-Type: application/json" \
  -d '{"jsonrpc":"2.0","id":1,"method":"tools/call","params":{"name":"hello","arguments":{}}}'

Installation

npm install

Configuration

Environment Variables

The MCP server uses environment variables for configuration:

• MIDDLEWARE_URL (required): Malaysia Transit Middleware API URL

  • Default: https://malaysiatransit.techmavie.digital

• INTERNAL_AUTH_TOKEN (required for production): Shared secret for middleware API authentication

  • Must match INTERNAL_APP_AUTH_TOKEN in the middleware's environment
  • Generate with: openssl rand -hex 32

• 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

• ANALYTICS_RESET_KEY (optional): Admin key for POST /analytics/reset

  • If not set, the reset endpoint is disabled (returns 403)

• ANALYTICS_IMPORT_KEY (optional): Admin key for POST /analytics/import

  • If not set, the import endpoint is disabled (returns 403)

Development

To run the MCP server in development mode:

npm run dev

Build

To build the MCP server for deployment:

npm run build

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:

const areas = await tools.list_service_areas();

get_area_info

Get detailed information about a specific area.

Parameters:

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

Example:

const info = await tools.get_area_info({ areaId: "penang" });

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:

const result = await tools.detect_location_area({ location: "KTM Alor Setar" });
// Returns: { area: "alor-setar", confidence: "high" }

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:

const stops = await tools.search_stops({
  area: "penang",
  query: "Komtar"
});

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:

const arrivals = await tools.get_stop_arrivals({
  area: "penang",
  stopId: "stop_123"
});
// Returns disclaimer + arrival data with confidence levels

find_nearby_stops

Find nearby stops and the routes that serve them.

Parameters:

• area (string): Service area ID
• location (string, optional): Place name to geocode (e.g., "KLCC", "Komtar")
• lat (number, optional): Latitude coordinate (required if location not provided)
• lon (number, optional): Longitude coordinate (required if location not provided)
• radius (number, optional): Search radius in meters (default: 500)

find_nearby_stops_with_arrivals ⭐

Find nearby stops and get real-time arrivals in one call.

Parameters:

• area (string): Service area ID
• location (string, optional): Place name to geocode
• lat (number, optional): Latitude coordinate (required if location not provided)
• lon (number, optional): Longitude coordinate (required if location not provided)
• radius (number, optional): Search radius in meters (default: 500)
• routeFilter (string, optional): Filter arrivals by route number (e.g., "302", "101")

find_nearby_stops_with_routes

Find nearby stops and get all routes serving those stops in one call.

Parameters:

• area (string): Service area ID
• location (string, optional): Place name to geocode
• lat (number, optional): Latitude coordinate (required if location not provided)
• lon (number, optional): Longitude coordinate (required if location not provided)
• 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_stops

Get all stops served by a specific route.

Parameters:

• area (string): Service area ID
• routeId (string): Route ID from list_routes

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:

const vehicles = await tools.get_live_vehicles({ area: "penang" });

get_provider_status

Check provider operational status.

Parameters:

• area (string): Service area ID

Schedule Information (NEW)

get_route_departures

Get the next N departures for a specific route (both directions).

Parameters:

• area (string): Service area ID (e.g., "ipoh", "seremban", "penang")
• routeId (string): Route ID from list_routes
• count (number, optional): Number of departures to return (default: 5)

Example:

const departures = await tools.get_route_departures({
  area: "ipoh",
  routeId: "A32",
  count: 5
});

get_next_departure

Get the single next departure for a route in a specific direction.

Parameters:

• area (string): Service area ID
• routeId (string): Route ID from list_routes
• direction (enum, optional): 'outbound', 'inbound', or 'loop'

get_stop_routes

Get all routes serving a specific stop with their next departures.

Parameters:

• area (string): Service area ID
• stopId (string): Stop ID from search_stops
• count (number, optional): Number of departures per route (default: 3)

get_route_schedule

Get the complete daily schedule for a route.

Parameters:

• area (string): Service area ID
• routeId (string): Route ID from list_routes

get_route_origin

Get the origin stop name for a route in a specific direction.

Parameters:

• area (string): Service area ID
• routeId (string): Route ID from list_routes
• direction (enum, optional): 'outbound' or 'inbound'

get_route_status

Check if a route is currently operating based on its schedule.

Parameters:

• area (string): Service area ID
• routeId (string): Route ID from list_routes

Fare Calculator (NEW)

get_fare_routes

Get all routes available for fare calculation in a specific area.

Parameters:

• area (string): Service area ID (e.g., "ipoh", "seremban", "penang", "kangar")

Supported Areas: BAS.MY areas (Ipoh, Seremban, Kangar, Alor Setar, Kota Bharu, Kuala Terengganu, Melaka, Johor, Kuching) and Rapid Penang.

get_route_stops_for_fare

Get all stops on a route with their distances for fare calculation.

Parameters:

• area (string): Service area ID
• routeId (string): Route ID from get_fare_routes

calculate_fare ⭐

Calculate the bus fare between two stops on a route.

Parameters:

• area (string): Service area ID
• routeId (string): Route ID from get_fare_routes
• fromStop (string): Origin stop ID
• toStop (string): Destination stop ID

Returns: Adult, concession, and child fares in MYR with disclaimer.

Example:

const fare = await tools.calculate_fare({
  area: "ipoh",
  routeId: "A32",
  fromStop: "stop_001",
  toStop: "stop_015"
});
// Returns: { adult: "1.50", concession: "0.75", child: "FREE" }

calculate_journey_fare

Calculate the total fare for a multi-leg journey with bus transfers.

Parameters:

• area (string): Base service area ID
• legs (array): Array of journey legs (max 5), each with:
  • routeId (string): Route ID for this leg
  • fromStop (string): Origin stop ID
  • toStop (string): Destination stop ID
  • areaId (string, optional): Area ID for this leg (for inter-area journeys)

Note: Each bus change requires a separate fare payment (BAS.MY does not have integrated transfers).

get_route_directions_for_fare

Get available directions for a route when calculating fares.

Parameters:

• area (string): Service area ID
• routeId (string): Route ID from get_fare_routes

KTM Komuter Utara Tools (NEW)

get_ktm_komuter_stations

Get all 23 KTM Komuter Utara stations (Padang Besar - Butterworth - Ipoh line).

Parameters: None

Returns: Station codes, names, and coordinates.

Example:

const stations = await tools.get_ktm_komuter_stations();
// Returns: [{ code: "PB", name: "Padang Besar", ... }, ...]

calculate_ktm_komuter_fare ⭐

Calculate KTM Komuter Utara fare between two stations.

Parameters:

• from (string): Origin station code (e.g., "BU" for Butterworth)
• to (string): Destination station code (e.g., "IP" for Ipoh)

Example:

const fare = await tools.calculate_ktm_komuter_fare({
  from: "BU",
  to: "IP"
});
// Returns: { adult: "12.00", child: "6.00", currency: "MYR" }

get_ktm_komuter_fare_matrix

Get the full KTM Komuter Utara fare matrix showing fares between all station pairs.

Parameters: None

get_ktm_station_departures ⭐

Get departure times for a specific KTM station.

Parameters:

• stationName (string): Station name (e.g., "Butterworth", "Ipoh")
• type (enum): Schedule type - ktm-komuter-utara or ktm-intercity

Example:

const departures = await tools.get_ktm_station_departures({
  stationName: "Butterworth",
  type: "ktm-komuter-utara"
});

get_ktm_stations

Get all KTM stations for a specific schedule type.

Parameters:

• type (enum): ktm-komuter-utara or ktm-intercity

get_ktm_schedules

Get full KTM schedule data for a specific schedule type.

Parameters:

• type (enum): ktm-komuter-utara or ktm-intercity

find_nearby_ktm_stations

Find KTM stations near a specific location.

Parameters:

• lat (number): Latitude coordinate
• lon (number): Longitude coordinate
• radius (number, optional): Search radius in kilometers (default: 10)
• type (enum): ktm-komuter-utara or ktm-intercity

Penang Ferry Tools (NEW)

get_penang_ferry_overview ⭐

Get Penang Ferry service overview including route, operator, and general info.

Parameters: None

get_penang_ferry_schedule

Get the full Penang Ferry schedule with operating hours and frequency.

Parameters: None

get_penang_ferry_next_departure

Get the next upcoming Penang Ferry departures from both terminals.

Parameters: None

get_penang_ferry_terminals

Get Penang Ferry terminal information including locations and facilities.

Parameters: None

get_penang_ferry_fare

Get Penang Ferry fare information and payment methods.

Parameters: None

Example:

const fare = await tools.get_penang_ferry_fare();
// Returns: { adult: "1.40", child: "0.70", ... }

System Tools

get_system_health

Check the health status of the Malaysia Transit middleware service.

Parameters: None

get_debug_info

Get comprehensive debug information about the middleware service.

Parameters: None

Analytics Tools (NEW)

get_api_analytics

Get API usage analytics and statistics from the middleware.

Parameters:

• type (enum, optional): Type of analytics to retrieve:
  • summary (default): Overview with requests/hour, error rate, uptime
  • endpoints: Per-endpoint statistics
  • areas: Per-service-area statistics
  • cumulative: All-time totals
  • clients: App/website usage breakdown

Example:

// Get summary analytics
const summary = await tools.get_api_analytics({ type: "summary" });

// Get client usage (see which apps use the API)
const clients = await tools.get_api_analytics({ type: "clients" });

get_area_analytics

Get detailed API usage analytics for a specific service area.

Parameters:

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

Example:

const penangStats = await tools.get_area_analytics({ area: "penang" });

Testing

hello

Simple test tool to verify server is working.

Usage Examples

Find When Your Bus is Coming

// 1. Detect area from location
const areaResult = await tools.detect_location_area({
  location: "KTM Alor Setar"
});

// 2. Search for your stop
const stops = await tools.search_stops({
  area: areaResult.area,
  query: "KTM Alor Setar"
});

// 3. Get real-time arrivals
const arrivals = await tools.get_stop_arrivals({
  area: areaResult.area,
  stopId: stops[0].id
});
// Returns: "Bus K100(I) arrives in 1 minute, Bus K100(O) in 2 minutes"

Track Live Buses

// Get all live vehicles in Penang
const vehicles = await tools.get_live_vehicles({
  area: "penang"
});

// Filter by bus only
const buses = await tools.get_live_vehicles({
  area: "klang-valley",
  type: "bus"
});

Discover Routes

// List all routes in Klang Valley
const routes = await tools.list_routes({
  area: "klang-valley"
});

// Get detailed route information
const routeDetails = await tools.get_route_details({
  area: "klang-valley",
  routeId: "LRT-KJ"
});

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:

1. User asks: "When is the next bus at Komtar?"
2. AI uses: detect_location_area({ location: "Komtar" })
3. AI uses: search_stops({ area: "penang", query: "Komtar" })
4. AI uses: get_stop_arrivals({ area: "penang", stopId: "..." })
5. AI responds: "Bus T101 arrives in 5 minutes, Bus T201 in 12 minutes"

2. "Where is my bus right now?"

Users want to track their bus in real-time.

Workflow:

1. User asks: "Where is bus T101 right now?"
2. AI uses: detect_location_area({ location: "Penang" })
3. AI uses: get_live_vehicles({ area: "penang" })
4. AI filters for route T101
5. AI responds: "Bus T101 is currently at [location], heading towards Airport"

Tool Usage Patterns

Always Start with Location Detection

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

// User: "When is the next bus at KTM Alor Setar?"
const areaResult = await tools.detect_location_area({ 
  location: "KTM Alor Setar" 
});
// Returns: { area: "alor-setar", confidence: "high" }

Search Before Details

Always search for stops/routes before requesting details:

// ✅ CORRECT
const stops = await tools.search_stops({ area: "penang", query: "Komtar" });
const arrivals = await tools.get_stop_arrivals({ 
  area: "penang", 
  stopId: stops[0].id 
});

// ❌ WRONG - Don't guess stop IDs
const arrivals = await tools.get_stop_arrivals({ 
  area: "penang", 
  stopId: "random_id" 
});

Response Formatting

Arrival Times

Format arrival times in a user-friendly way:

// ✅ GOOD
"Bus T101 arrives in 5 minutes"
"Train LRT-KJ arrives in 2 minutes"
"Next bus: T201 in 12 minutes"

// ❌ BAD
"Arrival time: 2025-01-07T14:30:00Z"
"ETA: 1736258400000"

Multiple Arrivals

Present multiple arrivals clearly:

"Upcoming arrivals at Komtar:
• T101 → Airport: 5 minutes
• T201 → Bayan Lepas: 12 minutes
• T102 → Gurney: 18 minutes"

Error Handling

Provider Unavailable

try {
  const arrivals = await tools.get_stop_arrivals({ ... });
} catch (error) {
  // Check provider status
  const status = await tools.get_provider_status({ area: "penang" });
  
  if (status.providers[0].status !== "active") {
    "The transit provider is currently unavailable. 
     Please try again later or check the official transit app."
  }
}

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

KTM Services

Location to Area Mapping

The detect_location_area tool automatically maps common locations to service areas:

Deployment

Production Server

The MCP server is deployed at:

• Endpoint: https://mcp.techmavie.digital/malaysiatransit/mcp
• Health Check: https://mcp.techmavie.digital/malaysiatransit/health
• Analytics Dashboard: https://mcp.techmavie.digital/malaysiatransit/analytics/dashboard
• Analytics API: https://mcp.techmavie.digital/malaysiatransit/analytics
• Transport: Streamable HTTP

Analytics Dashboard

The MCP server includes a built-in analytics dashboard that tracks:

• Total requests and tool calls
• Tool usage distribution (doughnut chart)
• Hourly request trends (last 24 hours)
• Requests by endpoint (bar chart)
• Top clients by user agent
• Recent tool calls feed

The dashboard auto-refreshes every 30 seconds.

Self-Hosting

To deploy your own instance, see the deployment guide.

# Using Docker
docker compose up -d --build

# Or run directly
npm run build
npm run start:http

Auto-Deployment

This repository includes a GitHub Actions workflow for automatic VPS deployment. When you push to main, the server automatically redeploys.

To set up auto-deployment, add these secrets to your GitHub repository:

• VPS_HOST - Your VPS IP address
• VPS_USERNAME - SSH username (e.g., root)
• VPS_PORT - SSH port (e.g., 22)
• VPS_SSH_KEY - Private SSH key for authentication
• GOOGLE_MAPS_API_KEY - Google Maps API key for geocoding
• INTERNAL_AUTH_TOKEN - Shared auth token (must match middleware's INTERNAL_APP_AUTH_TOKEN)
• ANALYTICS_RESET_KEY (optional) - Enables /analytics/reset admin endpoint
• ANALYTICS_IMPORT_KEY (optional) - Enables /analytics/import admin endpoint

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. Verify INTERNAL_AUTH_TOKEN matches middleware INTERNAL_APP_AUTH_TOKEN (if middleware auth is enabled)
5. 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

malaysiatransit-mcp/
├── src/
│   ├── index.ts              # Main MCP server entry point
│   ├── http-server.ts        # Streamable HTTP server with analytics
│   ├── transit.tools.ts      # Transit tool implementations
│   ├── geocoding.utils.ts    # Location detection utilities
│   ├── inspector.ts          # MCP Inspector entry point
│   └── server.ts             # HTTP server for testing
├── deploy/
│   ├── DEPLOYMENT.md         # VPS deployment guide
│   └── nginx-mcp.conf        # Nginx reverse proxy config
├── .github/
│   └── workflows/
│       └── deploy-vps.yml    # GitHub Actions auto-deploy
├── docker-compose.yml        # Docker deployment config
├── Dockerfile                # Container build config
├── package.json              # Project dependencies
├── tsconfig.json             # TypeScript configuration
├── .env.sample               # Environment variables template
├── README.md                 # This file
└── LICENSE                   # MIT License

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
• Model Context Protocol for the MCP framework
• Google Maps Platform for geocoding services
• OpenStreetMap Nominatim for fallback geocoding

Made with ❤️ by Aliff