Home Assistant MCP

Active
GitHub Python MIT

Description

The Unofficial and Awesome Home Assistant MCP Server — enables AI assistants to interact with smart home systems via the Model Context Protocol for intelligent device control and automation.

Key Features

  • 84+ tools for controlling smart home devices, querying states, executing services, and managing automations via natural language
  • Multi-platform setup — macOS, Linux, Windows, and Home Assistant OS add-on with guided installers
  • Supports 15+ AI clients including Claude Desktop, Claude Code, Gemini CLI, ChatGPT, VSCode, and Cursor
  • Remote access via Nabu Casa webhook proxy or Cloudflare Tunnel for secure outside-home control
  • Built with FastMCP framework — fast, lightweight Python MCP server with E2E test coverage
  • Demo environment included for testing without your own Home Assistant instance

Use Cases

💡 Controlling smart home devices and automations through natural language conversation with AI assistants
💡 Debugging and troubleshooting Home Assistant automations using AI-powered analysis
💡 Creating and modifying Lovelace dashboards through conversational UI
💡 Managing home security systems, lighting, climate, and entertainment via voice or text
💡 Building custom smart home routines and automations without writing YAML manually

Tags

python mcp tools automation api

Categories

🔌 MCP Tools

Quick Start

1. On macOS: run `curl -LsSf https://raw.githubusercontent.com/homeassistant-ai/ha-mcp/master/scripts/install-macos.sh | sh`
2. Download Claude Desktop from claude.ai/download (or restart if already installed)
3. Open Claude and ask: 'Can you see my Home Assistant?'
4. You're connected to the demo environment — go to Settings to connect your own HA instance
5. Try: 'Turn on the living room lights' or 'Create an automation for sunset'
Visit GitHub

Related Projects