Homey MCP Server
Enables AI agents to control Homey smart home devices, trigger flows, and query zones via the Homey API.
README
Homey MCP Server
This is an MCP server for Homey (by Athom) that allows AI agents to control smart home devices, trigger flows, and query zones.
Prerequisites
- Homey Pro or Homey Cloud
- Homey Personal Access Token (API Key). You can create one in the Homey Web App under
Settings → API Keys.
Installation
- Clone or copy this directory.
- Run
npm install. - Create a
.envfile based on.env.example:HOMEY_TOKEN=your_token_here
Usage
Connecting with Claude Desktop
Add the following to your claude_desktop_config.json:
{
"mcpServers": {
"homey": {
"command": "node",
"args": ["/Users/XX/.gemini/antigravity/scratch/homey-mcp/index.js"],
"env": {
"HOMEY_TOKEN": "YOUR_TOKEN_HERE"
}
}
}
}
Connecting with ChatGPT Desktop
ChatGPT (macOS/Windows) supports MCP servers via Developer Mode.
-
Open ChatGPT Desktop.
-
Go to Settings → Apps → Advanced settings.
-
Enable Developer Mode.
-
Click Create app (or "Add App") and configure a connection:
- Name: Homey
- URL:
http://localhost:3000(if running locally as an HTTP/SSE server) or use the Claude-style local command if supported in your version.
[!NOTE] ChatGPT currently primarily supports MCP over SSE (HTTP). To use this as a local command-line tool like in Claude, you may need a bridge or for ChatGPT to fully roll out stdio support.
Connecting with ChatGPT (Web/Custom GPTs)
To use Homey MCP in the ChatGPT web interface:
- You must host this MCP server on a public URL (e.g., via
ngrokor a VPS). - Use the Actions feature in a Custom GPT to connect to your MCP server's OpenAPI spec or SSE endpoint.
Available Tools (19 Total)
🏠 Devices & Control
homey_list_devices: List all devices with their current states (ID, Name, Zone, Class, Capabilities).homey_get_device: Get full details for a single device by its ID.homey_get_sensor_readings: Get current readings from all sensors (temperature, humidity, motion, etc.) across the home.homey_find_devices_by_zone: List all devices in a specific room or floor.homey_control_lights_in_zone: Bulk control for all lights in a specific zone (Turn on/off).homey_set_capability: Set a specific capability value (e.g.,onoff,dim,target_temperature).
🚀 Flows & Automation
homey_list_flows: List all standard flows.homey_list_advanced_flows: List all Advanced Flows.homey_run_flow: Trigger a standard flow by its ID.homey_run_advanced_flow: Trigger an Advanced Flow by its ID.homey_get_flow: Get full configuration details for a specific flow.homey_get_flow_folders: List the folder structure used to organize flows.
📍 Zones
homey_list_zones: List all zones (rooms/floors) and their hierarchy.
🧠 Logic & Flow Cards
homey_get_flow_cards: List available flow cards (triggers, conditions, actions) on the system.homey_run_flow_card_action: Execute a specific flow card action directly without needing a flow.homey_get_device_flow_capabilities: Identify which flow cards are applicable to a specific device.
📊 Data & Insights
homey_get_energy_data: Discover and list available energy/power logs.homey_get_device_insights: Fetch historical data logs for any device capability over a period (today,last7days, etc.).homey_get_live_insights: Access real-time dashboard data for device capabilities.
Troubleshooting
"HomeyAPI.createCloudAPI is not a function"
This usually happens if the Personal Access Token is used without a HOMEY_IP. Ensure you provide HOMEY_IP for Homey Pro local connections. For Homey Cloud, Ensure your token is valid and specifically created for the Web API.
Connection Closed / MCP Error
If the server fails to connect to Homey, it will now start gracefully but return an error message when you try to use a tool. Check the console logs (stderr) for the specific connection error.
Environment Variables
If running via Claude Desktop, ensure the env section in your config contains the correct HOMEY_TOKEN and optionally HOMEY_IP.
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。
mcp-server-qdrant
这个仓库展示了如何为向量搜索引擎 Qdrant 创建一个 MCP (Managed Control Plane) 服务器的示例。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。