ILP Drone Delivery MCP Server

Model Context Protocol server enabling Large Language Models to interact with the ILP Drone Delivery System through natural language

Overview

This MCP server allows AI assistants like Claude to plan drone deliveries, check availability, and visualize routes using natural language queries instead of manual API calls.

Example usage:

User: "Can you plan a delivery to Edinburgh Castle with 5kg capacity?"
Claude: [Uses MCP tools] "I can send Drone 3, estimated cost $12.50, 45 moves..."

Features

Available Tools

1. list_available_drones - Get all drones with capabilities
2. get_drone_details - Get specific drone information
3. plan_delivery - Plan a single delivery with cost/time estimates
4. check_drone_availability - Check which drones can handle requirements
5. get_delivery_geojson - Generate GeoJSON for map visualization
6. plan_multiple_deliveries - Plan multi-drone delivery routes

Prerequisites

• Node.js 18+ installed
• ILP CW2 Service running on http://localhost:8080
• Claude Desktop (for LLM integration) OR manual testing

🔧 Installation

Step 1: Set Up Project

cd ilp-mcp-server

# Install dependencies
npm install

# Make server executable
chmod +x server.js

# Link globally (for Claude Desktop)
npm link

Step 2: Start Your ILP Service

cd ILPCW2
java -jar target/*.jar app.jar

Verify it's running: curl http://localhost:8080/api/v1/dronesWithCooling/false

Step 3: Test the MCP Server

cd ilp-mcp-server
npm test

Expected output:

🧪 Testing ILP MCP Server

1️⃣  Testing API connection...
✅ Connected! Found 8 drones

2️⃣  Testing list_available_drones...
✅ Success! Retrieved 8 drones

3️⃣  Testing plan_delivery...
✅ Success! Planned delivery
   Cost: $11.06
   Moves: 26
   Drone: 1

4️⃣  Testing get_delivery_geojson...
✅ Success! Generated GeoJSON
   Type: FeatureCollection
   Features: 2

✅ All tests passed! (4/4)

🤖 Claude Desktop Integration

Configuration

Edit your Claude Desktop config file:

Mac: ~/Library/Application Support/Claude/claude_desktop_config.json

Windows: %APPDATA%\Claude\claude_desktop_config.json

Add this configuration:

{
  "mcpServers": {
    "ilp-drone": {
      "command": "node",
      "args": ["/Users/rheabose/ilp-mcp-server/server.js"]
    }
  }
}

💬 Example Queries

Try these in Claude Desktop:

Basic Queries

"What drones are available?"
"Show me drones with cooling capability"
"Get details for drone 3"

Planning Deliveries

"Plan a delivery to coordinates (-3.188, 55.945) with 4kg capacity"
"I need to deliver 5kg with heating to Edinburgh Castle"
"Can you plan a delivery to (lng: -3.19, lat: 55.94) requiring cooling?"

Checking Availability

"Which drones can handle a 6kg delivery with heating?"
"Check if any drones are available for a 3kg cooled delivery"

Visualization

"Generate a GeoJSON path for a delivery to (-3.188, 55.945) with 4kg capacity"
"Show me the route visualization for a delivery to Edinburgh"

Multi-Delivery

"Plan deliveries to these locations: 
 1. (-3.188, 55.945) - 4kg
 2. (-3.192, 55.943) - 3kg
 3. (-3.185, 55.946) - 5kg"

🧪 Manual Testing (Without Claude Desktop)

You can test the MCP server manually using the test script:

npm test

Or test individual API calls:

# Test list drones
curl http://localhost:8080/api/v1/dronesWithCooling/false

# Test plan delivery
curl -X POST http://localhost:8080/api/v1/calcDeliveryPath \
  -H "Content-Type: application/json" \
  -d '[{"id":999,"requirements":{"capacity":4.0},"delivery":{"lng":-3.188,"lat":55.945}}]'

🏗️ Architecture

┌─────────────────┐
│  Claude Desktop │
│   (LLM Client)  │
└────────┬────────┘
         │ MCP Protocol (stdio)
         │
┌────────▼────────┐
│   MCP Server    │
│   (server.js)   │
└────────┬────────┘
         │ HTTP REST API
         │
┌────────▼────────┐
│  ILP CW2 API    │
│  (Spring Boot)  │
└─────────────────┘

📝 Tool Descriptions

list_available_drones

• Purpose: Get all drones with capabilities
• Parameters:
  • hasCooling (optional): Filter by cooling capability
• Returns: List of drones with capacity, features, costs

plan_delivery

• Purpose: Plan a complete delivery route
• Parameters:
  • deliveryLng, deliveryLat: Delivery location
  • capacity: Required capacity in kg
  • heating, cooling (optional): Temperature requirements
  • date (optional): Delivery date
• Returns: Cost, moves, drone assignment, route summary

check_drone_availability

• Purpose: Find drones matching specific requirements
• Parameters:
  • capacity: Required capacity
  • heating, cooling (optional): Temperature needs
  • date (optional): Date to check
• Returns: List of available drone IDs

get_delivery_geojson

• Purpose: Generate map visualization data
• Parameters: Delivery location and requirements
• Returns: GeoJSON with flight paths

👤 Author

RheaBose University of Edinburgh - Informatics Large Practical