Home Assistant AI 控制服务

Home Assistant MCP Server & AI Control Service

A comprehensive solution for controlling Home Assistant with AI, featuring:

• MCP Server: Model Context Protocol server for Home Assistant integration
• Spring Boot Service: Web service with REST API, Vaadin Flow UI, and Spring AI + Ollama integration

Project Structure

homeassistant-mcp/
├── mcp-server/          # Java MCP server for Home Assistant
│   └── src/main/java/com/homeassistant/mcp/
│       ├── HomeAssistantMcpServer.java
│       └── HomeAssistantClient.java
│
└── spring-service/      # Spring Boot web service
    └── src/main/java/com/homeassistant/ai/
        ├── client/      # Home Assistant REST client
        ├── config/      # Configuration classes
        ├── controller/  # REST API controllers
        ├── service/     # AI chat service
        └── view/        # Vaadin Flow UI

Features

MCP Server

• Standard MCP protocol implementation
• Tools for Home Assistant operations:
  • get_states - Get all entity states
  • get_state - Get specific entity state
  • call_service - Execute Home Assistant services
  • get_services - List available services
  • get_config - Get Home Assistant configuration

Spring Boot Service

• REST API: Full Home Assistant control via HTTP endpoints
• Vaadin Flow UI: Interactive web interface with:
  • AI chat assistant powered by Ollama
  • Real-time device listing and control
  • Turn on/off lights, switches, fans, etc.
• Spring AI Integration: Chat with AI about your home and control devices

Prerequisites

• Java 21 (Amazon Corretto 23 JDK)
• Maven 3.8+
• Home Assistant instance running at http://homeassistant.local:8123
• Ollama running locally at http://localhost:11434
• Ollama model installed (e.g., ollama pull llama3.2)

Configuration

Home Assistant Access Token

The projects are pre-configured with your Home Assistant access token. To change it:

MCP Server: Edit mcp-server/src/main/java/com/homeassistant/mcp/HomeAssistantMcpServer.java or set environment variable:

export HA_ACCESS_TOKEN="your-token-here"
export HA_BASE_URL="http://homeassistant.local:8123"

Spring Service: Edit spring-service/src/main/resources/application.properties:

homeassistant.base-url=http://homeassistant.local:8123
homeassistant.access-token=your-token-here

Ollama Configuration

Configure Ollama settings in spring-service/src/main/resources/application.properties:

spring.ai.ollama.base-url=http://localhost:11434
spring.ai.ollama.chat.options.model=llama3.2
spring.ai.ollama.chat.options.temperature=0.7

Building & Running

MCP Server

Build:

cd mcp-server
mvn clean package

Run:

java -jar target/homeassistant-mcp-server-1.0.0.jar

The MCP server communicates via stdin/stdout following the MCP protocol.

Spring Boot Service

Build:

cd spring-service
mvn clean package

Run:

mvn spring-boot:run

Or run the JAR:

java -jar target/homeassistant-ai-service-1.0.0.jar

Access the application:

• Web UI: http://localhost:8080
• REST API: http://localhost:8080/api/

REST API Endpoints

Home Assistant Control

• GET /api/homeassistant/states - Get all entity states
• GET /api/homeassistant/states/{entityId} - Get specific entity state
• POST /api/homeassistant/services/{domain}/{service} - Call a service
• GET /api/homeassistant/services - List all services
• GET /api/homeassistant/config - Get configuration

AI Chat

• POST /api/chat - Send a message to the AI
  复制

  {
    "message": "Turn on the living room lights"
  }
  

• POST /api/chat/clear - Clear conversation history

Usage Examples

Using the Web UI

1. Navigate to http://localhost:8080
2. Chat with the AI assistant about your home
3. Control devices directly with the on/off buttons
4. Ask questions like:
   • "What lights are currently on?"
   • "Turn off all the lights in the bedroom"
   • "What's the temperature?"

Using the REST API

Get all states:

curl http://localhost:8080/api/homeassistant/states

Turn on a light:

curl -X POST http://localhost:8080/api/homeassistant/services/light/turn_on \
  -H "Content-Type: application/json" \
  -d '{"entity_id": "light.living_room"}'

Chat with AI:

curl -X POST http://localhost:8080/api/chat \
  -H "Content-Type: application/json" \
  -d '{"message": "What devices are on right now?"}'

Using the MCP Server

The MCP server follows the Model Context Protocol. Example initialization:

{
  "jsonrpc": "2.0",
  "id": 1,
  "method": "initialize",
  "params": {
    "protocolVersion": "2024-11-05",
    "capabilities": {}
  }
}

Development

Adding New Features

• MCP Server: Add new tools in HomeAssistantMcpServer.handleToolsList()
• REST API: Create new controllers in spring-service/src/main/java/com/homeassistant/ai/controller/
• UI: Modify MainView.java or add new Vaadin views

Testing

Test Home Assistant connection:

curl -H "Authorization: Bearer YOUR_TOKEN" \
  http://homeassistant.local:8123/api/states

Test Ollama:

curl http://localhost:11434/api/generate -d '{
  "model": "llama3.2",
  "prompt": "Hello, world!"
}'

Troubleshooting

Cannot connect to Home Assistant:

• Verify Home Assistant is running
• Check the URL and access token
• Ensure the token has the necessary permissions

Cannot connect to Ollama:

• Start Ollama: ollama serve
• Verify the model is installed: ollama list
• Pull the model if needed: ollama pull llama3.2

Vaadin compilation issues:

• Run mvn vaadin:prepare-frontend
• Clear npm cache if needed: rm -rf node_modules

License

MIT License - feel free to modify and use as needed.

Contributing

Contributions are welcome! Feel free to submit issues or pull requests.