mcpField
H
Home Assistant Model Context Protocol (MCP)

Home Assistant Model Context Protocol (MCP)

Local
jango-blockchained.github.ioAI & Agents58Updated 19w ago

About Home Assistant Model Context Protocol (MCP)

An MCP server that connects AI assistants to Home Assistant, offering over 50 tools to control lights, climate, locks, scenes and automations. It suits users who want to manage a smart home from clients such as Claude Desktop or Cursor. It runs over STDIO and HTTP transports.

Key features

  1. 01Over 50 tools covering smart home devices
  2. 02Supports STDIO, HTTP+WS and HTTP transports
  3. 03Offers wake word detection and speech-to-text
  4. 04Installable via Docker, npx, bunx or npm

Use cases

  1. 01Controlling lights, thermostats and locks through an AI assistant
  2. 02Querying historical states and energy consumption
  3. 03Managing home automations from Cursor or Claude Desktop

MCP support

The project is an MCP server that exposes Home Assistant device control, state queries and automation operations as tools to MCP clients.

Configuration

Local server config is the same for every client. Copy it into your client’s MCP config file.

mcp.json · Universal config
{
"mcpServers": {
"advanced-homeassistant": {
"command": "npx",
"args": [
"-y",
"@smithery/cli"
]
}
}
}
Run commandnpx -y @smithery/cli

Local servers need a runtime on your machine (e.g. Node.js 18+). Config file locations differ by client; see your client’s docs.

Packages
npm@smithery/cli

Tools

No tool list yet

Local servers expose tools only once running on your machine, and the maintainer has not provided a tool list.

Details

From the project README

🏠 Home Assistant MCP

Smithery Release npm Docker CI Bun Node License

Talk to your house. Your AI assistant (Claude, GPT, Cursor, Copilot) — connected to Home Assistant through the Model Context Protocol. 50+ tools, three transports, one bunx command.

# ⚡ Fastest way to try it
bunx github:jango-blockchained/advanced-homeassistant-mcp

🚀 Quick Start

You need a Home Assistant instance and a long-lived access token (create one here).

Claude Desktop

Add to claude_desktop_config.json:

{
  "mcpServers": {
    "homeassistant-mcp": {
      "command": "bunx",
      "args": ["github:jango-blockchained/advanced-homeassistant-mcp"],
      "env": {
        "HASS_HOST": "http://your-ha-instance:8123",
        "HASS_TOKEN": "your_long_lived_access_token"
      }
    }
  }
}

Restart Claude. You're done. Ask it to turn on the lights.

Cursor / VS Code
Cursor — add to .cursor/mcp.json
{
  "mcpServers": {
    "homeassistant-mcp": {
      "command": "bunx",
      "args": ["github:jango-blockchained/advanced-homeassistant-mcp"],
      "env": {
        "HASS_HOST": "http://your-ha-instance:8123",
        "HASS_TOKEN": "your_long_lived_access_token"
      }
    }
  }
}
VS Code + MCP Extension

Set your env vars in a .env file and use the pre-configured .vscode/mcp.json:

{
  "mcp.servers": {
    "homeassistant-mcp": {
      "type": "stdio",
      "command": "bun",
      "args": ["run", "dist/stdio-server.mjs"],
      "env": {
        "HASS_HOST": "${env:HASS_HOST}",
        "HASS_TOKEN": "${env:HASS_TOKEN}"
      }
    }
  }
}
Smithery (One-Click)

Install directly from the Smithery registry — no config files needed:

npx @smithery/cli install @jango-blockchained/advanced-homeassistant-mcp --client claude

📦 All Install Options

MethodCommand
Smithery (easiest)npx @smithery/cli install @jango-blockchained/advanced-homeassistant-mcp --client claude
bunx (no install)bunx github:jango-blockchained/advanced-homeassistant-mcp
npxnpx @jango-blockchained/homeassistant-mcp@latest
Dockerdocker run -d -p 7123:7123 --env-file .env ghcr.io/jango-blockchained/advanced-homeassistant-mcp:1.7.0
From sourcegit clone ... && bun install && bun run build:all && bun run start:stdio
npm globalnpm add -g @jango-blockchained/homeassistant-mcp && homeassistant-mcp

Available Docker tags:

  • latest - Latest stable release
  • 1.7.3, 1.7, 1 - Tagged versions
  • dev - Latest development build from main branch

🎯 What You Can Do

Once connected, you talk to your house like you talk to a human:

"Turn off all lights in the bedroom"
"Set the thermostat to 72°F"
"Lock all doors and start the vacuum"
"Show me energy consumption this week"
"What's the temperature in the living room?"
"Activate the movie scene"
"Notify everyone that dinner is ready"
"Check my Home Assistant health"
"Find orphaned devices with low battery"
"Analyze my light usage patterns"

Every command maps to one of 50+ tools — lights, climate, media, locks, covers, fans, vacuums, alarms, scenes, automations, notifications, history, energy monitoring, maintenance, and more.

Full tool reference (all 50+ tools)
CategoryTools
Lightslist, get, turn on/off, brightness, color temp, RGB, effects
Climatelist, get, HVAC modes, target temp, fan mode, humidify
Medialist, get, play/pause, volume, source, sound mode, shuffle
Coverslist, get, open/close, position, tilt, garage door
Lockslist, get, lock/unlock with code support
Fanslist, get, speed, oscillation, direction, preset
Vacuumslist, get, start/stop/dock, spot clean, fan speed
Alarmlist, get, arm home/away/night, disarm
Switcheslist, get, turn on/off/toggle
Sceneslist, activate named scene
Automationslist, toggle, trigger, create/edit/delete
Notificationspush alerts via HA channels
Historyquery historical states
Add-onsinstall, configure, control
PackagesHACS integrations and custom components
Maintenanceorphaned devices, battery warnings, energy analysis
Smart Scenariosnobody-home mode, window/heat conflicts, energy waste
Lightinganimations, scenarios, showcase, BPM/beat detection
Voicewake word detection, speech-to-text, voice commands
Dashboardquery and manage Lovelace dashboards
Templatesrender Jinja2 templates via HA
To-Do Listsadd, update, remove items
Tracesautomation/script execution traces
Searchfull-text entity search with filters
Entity Stateget current state of any entity
Error Logquery HA error logs with filtering

Full docs: Tools Reference


🎙️ Speech Features (Voice Control)

Transform your setup into a fully voice-controlled smart home assistant. The speech integration provides local, privacy-focused voice processing.

  • Wake Word Detection: Powered by wyoming-openwakeword (default: "Hey Jarvis").
  • Speech-to-Text: Fast, local transcription using faster-whisper.
  • Audio Integration: Direct PulseAudio integration for seamless microphone access.

To enable speech features, use the dedicated Docker Compose file:

docker-compose -f docker-compose.speech.yml up -d

See the Speech Features Guide for detailed configuration and setup instructions.


🧱 Architecture

┌──────────────┐     ┌─────────────────────┐     ┌──────────────────┐
│  AI Assistant │◄───►│   Home Assistant    │◄───►│  Your Smart Home │
│ (Claude/GPT)  │     │    MCP Server       │     │  (Home Assistant)│
└──────────────┘     └─────────────────────┘     └──────────────────┘
                         │           │
                    ┌────┘           └────┐
                    ▼                     ▼
              ┌────────────┐      ┌──────────────┐
              │  HTTP/WS   │      │    STDIO     │
              │ (Express)  │      │ (fastmcp v3) │
              └────────────┘      └──────────────┘

Three transports, one codebase. Pick the one that fits your setup — all expose the same 50+ tools.

  • STDIO — for Claude Desktop, Cursor, VS Code (local editor integrations)
  • HTTP+WS — for remote AI hosts, with JWT auth, rate limiting, and WebSocket streaming
  • HTTP (fastmcp) — lightweight HTTP transport without bespoke middleware

⚙️ Configuration

Minimal setup. Two variables are required:

HASS_HOST=http://homeassistant.local:8123
HASS_TOKEN=eyJ...                          # long-lived access token

Optional:

PORT=7123                                  # HTTP server port (default: 4000)
LOG_LEVEL=info                             # debug | info | warn | error
JWT_SECRET=...                             # for HTTP/WS auth (min 32 chars)

For the full list of 20+ environment variables, see the Configuration docs.


🛠️ Development

bun install              # install dependencies
bun run build:all        # build all three entry points
bun test                 # run test suite (80% coverage threshold)
bun run lint             # ESLint + Prettier
bun run typecheck        # TypeScript type checking

Built with Bun, TypeScript, and fastmcp. Three entry points:

  • dist/index.cjs — HTTP+WS on a custom Express-based MCP server
  • dist/stdio-server.mjs — STDIO via fastmcp v3
  • dist/http-server.mjs — HTTP via fastmcp v3

📖 Docs

Full documentation lives at jango-blockchained.github.io/advanced-homeassistant-mcp — covers installation, configuration, all 50+ tools, deployment (HTTP, STDIO, Smithery), architecture deep-dives, and guides.


🤝 Contributing

PRs welcome. Keep it simple — same code style, add tests, update docs if you touch public APIs.

  1. Fork it
  2. Branch it
  3. Code it
  4. Test it
  5. PR it

See CONTRIBUTING.md for the full guidelines.


📄 License

MIT — go build something cool.


Your smart home speaks MCP. Now your AI speaks it too. 🏠⚡🤖🔋

Batteries included. 🔋

Comments

Markdown supported

New to MCP?

From what the protocol is to installing your first server, these answers cover most first questions.

An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.