Spec Driven Development
MCP Server that facilitates spec-driven development workflows, not just Vibe Coding.
Descripción del creador (original en inglés)
LAN-first, cloud-fallback MCP server for Govee smart lights and devices.
Estos tres nunca se combinan en una sola puntuación: las opiniones, las mediciones y la popularidad son cosas distintas.
Todavía no escribimos un resumen sencillo para este. Abajo están los datos que leímos del registro.
Elige la app de IA que usas y copia la configuración.
GOVEE_API_KEYLos valores marcados como secretos son llaves de tu cuenta. Nunca los compartas ni los pegues en un chat.
Necesitas
¿Qué app usas?
Elige tu app arriba. La recordamos, así que la próxima vez el botón dirá “Instalar en …”.
Abre Claude Desktop → Settings → Developer → Edit Config y pega esto en claude_desktop_config.json
{
"mcpServers": {
"govee": {
"command": "uvx",
"args": [
"govee-mcp"
],
"env": {
"GOVEE_API_KEY": "your-value"
}
}
}
}Ejecuta esto en tu terminal
claude mcp add govee --env GOVEE_API_KEY=your-value -- uvx govee-mcpCompleta los valores vacíos antes de ejecutarlo.
Pégalo en .cursor/mcp.json de tu proyecto, o en ~/.cursor/mcp.json
{
"mcpServers": {
"govee": {
"command": "uvx",
"args": [
"govee-mcp"
],
"env": {
"GOVEE_API_KEY": "your-value"
}
}
}
}Tu navegador te pide permiso para abrir la app; luego la app muestra la configuración y te pide confirmar. Los valores de las claves quedan vacíos para que los completes tú.
Pégalo en .vscode/mcp.json de tu proyecto
{
"servers": {
"govee": {
"type": "stdio",
"command": "uvx",
"args": [
"govee-mcp"
],
"env": {
"GOVEE_API_KEY": "your-value"
}
}
}
}Tu navegador te pide permiso para abrir la app; luego la app muestra la configuración y te pide confirmar. Los valores de las claves quedan vacíos para que los completes tú.
Este se ejecuta en tu computadora, y ChatGPT solo puede conectarse a servidores en internet. Elige otra app.
Tu elección se guarda en una cookie en este dispositivo; no necesitas cuenta.
Esta configuración se genera a partir de la ficha del creador en el registro. Si el README dice otra cosa, sigue el README.
Se ejecuta en tu computadora; la lista de herramientas aparece en tu app de IA después de instalarlo.
Revisamos información que cualquiera puede ver; no hemos leído el código. Las revisiones que no aplican no se cuentan. · Revisado el 9 nov 2026
El nombre en el registro se verificó con la cuenta de GitHub evefromwayback, dueña del repo
Todavía no hemos leído este repo en GitHub
Todavía no hemos leído este repo en GitHub
Este se ejecuta en tu computadora; no hay endpoint en internet
Se ejecuta en tu computadora y usa ajustes locales en su lugar
Se ejecuta en tu computadora; solo podemos leer listas de herramientas de endpoints en internet
Comparado con todos los servidores activos de esta categoría en el registro. Las estrellas solo cuentan cuando quien lo publica es dueño del repo.
Superó 1 de 1 revisiones básicas aplicables. No hemos leído el código. Antes de agregar una clave o un token, revisa qué acceso pide el servidor y usa una clave que tenga solo los permisos que necesitas.
Debes completar: GOVEE_API_KEY
Se ejecuta en tu computadora. Tu app de IA lo inicia cuando hace falta, y los datos no pasan por los servidores del creador (a menos que el propio servidor llame a un servicio en línea).
El repo de GitHub cambió por última vez –, y la última versión (0.2.0) llegó al registro hace 5 meses.
Las apps compatibles con MCP, como Claude Desktop, Claude Code, Cursor y VS Code. También necesitas Python (uv) en tu computadora.
En la misma categoría: Spec Driven Development, Remote Desktop Commander, Shield Kya
Servidores de la misma categoría hechos por otros creadores, ordenados por estrellas.
Cómo leer una tarjeta:↓ Descargas por semanai★ Estrellas de GitHubiTres pilares: reseñas · revisiones · usoiActualizado = fecha en que se publicó la última versión en el registroi
MCP Server that facilitates spec-driven development workflows, not just Vibe Coding.
Remote Desktop Commander: use files, terminal and processes on your own computers from any AI app.
Audit trail and traceability for AI agents. MCP gate: evaluate, ingest, request approval.
Copiado tal cual del repo del creador. PopNx MCP no lo ha revisado ni lo respalda. Se quitaron HTML, imágenes y scripts. Los enlaces llevan fuera de este sitio.
A LAN-first, cloud-fallback Model Context Protocol (MCP) server for Govee smart devices. Point an AI agent, or any MCP client, at your Govee lights, plugs, and appliances and control them by name.
It is built to run on a home network. It prefers local LAN control for lights (fast, no internet, no rate limits) and falls back to Govee's cloud API for plugs, appliances, and anything not reachable on the LAN.
This is a v0 that works for me on my own network. It is shared in the spirit of "here is something that works with an agent on my home network, have it, make it yours, and harden it as you please." It is not a supported product. Issues and pull requests are welcome, but treat it as a starting point rather than a finished appliance.
These are deliberately left as room to grow.
The existing Govee MCP servers are thin: one is cloud-only and controls a single hardcoded device with three tools; another is LAN-only and lights-only. This one does multi-device discovery, both transports with automatic selection, state read-back, and cloud rate-limit handling.
From source:
git clone https://github.com/evefromwayback/govee-mcp.git
cd govee-mcp pip install -e .
This installs a govee-mcp command. To run without installing:
PYTHONPATH=src python3 -m govee_mcp
Copy .env.example to .env and fill in what you need, or set the variables in your MCP client config.
| Variable | Default | Purpose |
|---|---|---|
GOVEE_API_KEY | (none) | Cloud API key. Without it, the server runs LAN-only. |
GOVEE_PREFER_LAN | true | Prefer LAN over cloud when a device is reachable both ways. |
GOVEE_LAN_ENABLED | true | Enable the LAN transport. |
GOVEE_CLOUD_ENABLED | true | Enable the cloud transport (also needs an API key). |
GOVEE_LAN_SCAN_IPS | (none) | Comma-separated device IPs to scan directly when multicast is flaky. |
GOVEE_LAN_DISCOVERY_TIMEOUT | 2 | Seconds to wait for LAN discovery replies. |
GOVEE_REQUEST_TIMEOUT | 10 | Cloud HTTP timeout in seconds. |
GOVEE_DEVICE_CACHE_TTL | 86400 | Seconds to cache the device list before re-discovering. Run govee refresh to update sooner. |
GOVEE_CACHE_ENABLED | true | Cache the device list on disk so most commands skip discovery. |
GOVEE_CACHE_PATH | ~/.cache/govee-mcp/devices.json | Where the on-disk device cache is written. |
A small CLI ships alongside the server and uses the same engine, so it is the fastest way to confirm things work before wiring up an MCP client. Start with doctor.
govee doctor # check key, cloud, and LAN connectivity
govee devices # list devices (instant, from cache) govee refresh # re-scan devices and show current state govee state # full current state for every device govee state "Desk Lamp" # full current state for one device govee on "Desk Lamp" govee off "Desk Lamp" govee brightness "Desk Lamp" 40 govee color "Desk Lamp" 255 0 0 govee temp "Desk Lamp" 4000 govee scenes "Desk Lamp" # list this device's scenes govee scene "Desk Lamp" "Sunrise" govee groups # list device groups govee group "Living Room" "Bulb 1" "Bulb 2" # create or replace a group govee ungroup "Living Room" # delete a group
Run govee help, or just govee with no arguments, for the command reference. doctor reports whether your key loaded, how many devices are on your cloud account, and how many answered on the LAN, with a hint for whatever is missing. If govee is not on your PATH, use python3 -m govee_mcp.cli ... instead.
Any device name in the control commands can also be a group name. Define a group with govee group "Living Room" "Bulb 1" "Bulb 2", then govee on "Living Room", govee color "Living Room" 255 0 0, or govee state "Living Room" apply to every device in it. Groups are stored in ~/.config/govee-mcp/groups.json (override with GOVEE_GROUPS_PATH).
govee devices is instant: it lists devices from the cache and makes no live calls. govee refresh re-scans your devices and then prints the live state of all of them. govee state (with or without a name) is always a live read.
Claude Desktop (claude_desktop_config.json), installed command:
{
"mcpServers": { "govee": { "command": "govee-mcp", "env": { "GOVEE_API_KEY": "your-key-here" } } } }
Running from source instead:
{
"mcpServers": { "govee": { "command": "python3", "args": ["-m", "govee_mcp"], "env": { "GOVEE_API_KEY": "your-key-here", "PYTHONPATH": "/absolute/path/to/govee-mcp/src" } } } }
Hermes Agent has a native MCP client, so this server drops in. Once it is in the Hermes catalog, install it by name:
hermes mcp install govee
To wire it manually, add it under mcp_servers in your Hermes config and put GOVEE_API_KEY in your Hermes .env (the server process inherits it):
mcp_servers:
govee: command: govee-mcp
If govee-mcp is not on the launch PATH, use its absolute path or python3 -m govee_mcp. A Hermes catalog manifest is included at hermes/manifest.yaml.
list_devices(): discovered devices with names, supported actions, and transports. Call this first.list_groups(): device groups and their members.create_group(name, devices): create or replace a group.delete_group(name): delete a group.get_device_state(name): current power, brightness, and color.refresh(): re-scan devices and return the live state of all of them.set_power(name, on): turn on or off.set_brightness(name, level): 0 to 100.set_color(name, red, green, blue): each channel 0 to 255.set_color_temp(name, kelvin): roughly 2000 (warm) to 9000 (cool).list_scenes(name): the dynamic scenes (preset effects) a device supports.set_scene(name, scene): apply a scene by name. Call list_scenes first to get valid names.Every control tool's name can be a single device or a group name (see list_groups); a group applies the action to all its members.
Scenes (the "patterns": Sunrise, Aurora, and so on) are cloud-only and differ per device, so they need an API key and are fetched live from each device rather than hardcoded. The other actions work over both transports. Each control tool returns a structured result. A change is only real when the tool returns ok=true; the tool descriptions instruct the model not to claim a change happened without a successful call.
On each action the registry picks a transport:
Set GOVEE_PREFER_LAN=false to always use the cloud, or disable a transport entirely with GOVEE_LAN_ENABLED=false or GOVEE_CLOUD_ENABLED=false.
Govee's cloud Developer API is free but, per Govee, for non-commercial use only. You may not use it for profit-making purposes or charge anyone for an implementation built on it. That restriction is on the cloud service and applies to whoever supplies the key, not to this code. Personal home use and sharing this project for free are both fine. Do not build a paid product on the free key without arranging commercial terms with Govee. See Govee's API service terms: https://app-h5.govee.com/common/api-service-item
You do not always need a key:
GOVEE_CLOUD_ENABLED=false..env.pip install -e ".[test]"
pytest
The tests mock the cloud HTTP layer and run a local fake Govee device over UDP, so they need no API key and no real hardware.
MIT. See LICENSE.
The protocols were implemented from Govee's public LAN guide and Developer API v2 documentation. Thanks to the open-source projects that mapped this territory first (govee2mqtt and govee-led-wez, homebridge-govee, and the Govee BLE sensor projects). No code was copied from them; they are credited as references.