HueMCP
Enables discovery and control of Philips Hue lighting devices via a local bridge using the CLIP v2 API, without any cloud dependency.
README
HueMCP
An MCP server that lets Claude (or any MCP client) discover and control a Philips Hue bridge over its local CLIP v2 HTTP API. No cloud account needed — all traffic is LAN-local to your bridge.
Tool surface
| Tool | Purpose |
|---|---|
list_lights() |
Every light with on/off, brightness, colour, owner device |
list_rooms() |
Rooms with their grouped_light_id and member devices |
list_zones() |
Zones with their grouped_light_id and member lights |
list_scenes(group_id=None) |
Scenes, optionally filtered to one room/zone |
set_light(light_id, on?, brightness?, color_temp_mirek?, color_xy?, effect?, gradient_points?, gradient_mode?, timed_effect?, timed_effect_duration_ms?, transition_ms?) |
Update one light |
set_group(grouped_light_id, on?, brightness?, color_temp_mirek?, color_xy?, transition_ms?) |
Update all lights in a room or zone |
activate_scene(scene_id, brightness?, transition_ms?, action="active") |
Recall a scene |
identify_light(light_id) |
Briefly pulse a light to locate it |
bridge_info() |
Basic identity of the bridge |
get_resource(rtype, rid=None) |
Generic escape hatch — raw bridge payload for any resource type |
brightness is a percent (0–100). color_temp_mirek is a CIE mired value
(typical range 153–500 ≈ 6500K–2000K — the per-light valid range is reported
by list_lights()). color_xy is [x, y] in CIE 1931 space, clamped to the
light's gamut.
list_lights() also reports per-light capability: effects.available
(e.g. candle, fire, prism, sparkle, opal, glisten), timed_effects.available
(sunrise, sunset), and — for gradient lightstrips — a gradient block
with points_capable, pixel_count, and mode_values. Pass matching values
to set_light to drive them: effect="candle", gradient_points=[[x,y], ...],
timed_effect="sunrise", etc.
Requirements
- Python 3.10+
- A Hue bridge reachable on your LAN
- An application key — generated by pressing the link button and POSTing to
/api(see "Bridge-side setup" below) - An MCP-capable client — these instructions assume Claude Code
Install
git clone <your-fork-url> HueMCP
cd HueMCP
python3 -m venv .venv
.venv/bin/pip install -e .
cp .env.example .env
# edit .env to set HUE_BRIDGE_HOST and HUE_APPLICATION_KEY
Smoke test:
.venv/bin/python scripts/smoke_test.py
Register with Claude Code (user scope, available in every project):
claude mcp add huemcp /absolute/path/to/HueMCP/.venv/bin/huemcp -s user
Bridge-side setup
-
Find your bridge's IP at https://discovery.meethue.com or in the Hue app (Settings → My Hue System).
-
Press the bridge's physical link button.
-
Within 30 seconds, request an application key:
curl -sk -X POST https://<bridge-ip>/api \ -H 'Content-Type: application/json' \ -d '{"devicetype":"huemcp#'"$(hostname)"'", "generateclientkey":true}'Copy the
usernamevalue from the response intoHUE_APPLICATION_KEY.
TLS
The bridge presents a self-signed certificate signed by the Philips Hue root CA.
By default HUE_VERIFY_TLS=0 and the client skips verification — appropriate
for a LAN-local device. To verify properly, install the Hue root CA into your
trust store and set HUE_VERIFY_TLS=1, or pin a CA bundle by setting
HUE_VERIFY_TLS=/path/to/ca.pem.
Configuration reference
| Variable | Default | Purpose |
|---|---|---|
HUE_BRIDGE_HOST |
(required) | Hostname or IP of the bridge |
HUE_APPLICATION_KEY |
(required) | Application key from the bridge |
HUE_VERIFY_TLS |
0 |
0 skip, 1 system trust, or path to CA bundle |
HUE_TIMEOUT |
10 |
Per-request timeout in seconds |
License
MIT — see LICENSE.
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。
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。
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。
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。
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。
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。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
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。