Yeelight Metadata MCP Server

io.github.Yeelight由 GitHub 帳號 Yeelight 發布v1.0.0更新於 Oct 24, 2026

Yeelight home metadata and guarded management workflows.

已驗證Streamable HTTP可網頁執行位置與生活效率與工作流

概覽

AI 產生的概覽

Yeelight 雲端 MCP 伺服器,讓助理取得智慧家庭的结构化檢視,並以受保護的方式管理房間、裝置、情境與自動化。

可以做什麼
  • 探索任務群組、任務與執行階段動作目錄,並讀取每個動作的參數結構
  • 列出家庭、房間、裝置、群組、面板、情境、自動化、收藏與維護項目
  • 透過統一的 execute_task 工具規劃並執行家庭管理任務,預設只產生 dry-run 計畫
  • 讀取 MCP 資源與提示詞,用於任務探索、變更規劃與自動化疑難排解
適用情境
適用於支援 MCP 但無法安裝 Agent Skills 的 AI 用戶端,需要理解並管理 Yeelight 智慧家庭時。它偏重房間、裝置、群組、情境、自動化與面板管理,即時裝置控制由搭配的 IoT 服務負責。
執行需求
使用 Yeelight 雲端的遠端 Streamable HTTP 端點,無需本機安裝。必須在 Authorization 標頭提供 Yeelight 雲端 Bearer 權杖,另可選用 Yeelight-Region 與 House-Id 標頭。從原始碼執行需要 Python 3.10 以上版本。
安裝前請注意
Authorization 權杖等同完整帳號憑證:不要貼進 AI 對話,建議改用 QR 登入流程,讓權杖留在本機代理中。任務可建立、更新、刪除、解除綁定、轉移或登入,實際執行 S2/S3 操作需 confirmSideEffect=true;請先做 dry run 並使用可復原的資料。家庭資料會傳送到 Yeelight 雲端服務。

安裝

在 SourceWeft 中

  1. 開啟 儀表板中的 Yeelight Metadata MCP Server,將其新增到工作區。
  2. 連接 Authorization token 憑證。它們會被私密儲存,絕不會在公開頁面上展示。
  3. 為需要使用其工具的對話啟用該服務。

Web executable,透過 Streamable HTTP。 遠端服務在工作區中設定後即可從網頁執行環境執行。

交給代理程式安裝

把這段提示詞貼上到 Claude Code、Codex、Cursor 等代理程式,或 SourceWeft 對話中。代理程式會讀取該伺服器的安裝指南,先向你展示發布者、端點和登入方式,經你同意後再把它新增到你的客戶端。金鑰不會寫進任何命令。

閱讀 https://sourceweft.com/mcp/io.github.Yeelight%2Fyeelight-metadata-mcp/install.md,並按其中說明把這個 MCP 伺服器接入我的客戶端。修改任何設定之前,先告訴我它的發布者、端點和登入方式,等我確認。不要把 API 金鑰寫進命令或我會提交的檔案;問我金鑰應從哪裡取得。

檢視代理程式所依據的安裝指南

其他 MCP 客戶端

把它新增到你客戶端的 mcpServers 設定中。

{
  "mcpServers": {
    "yeelight-metadata": {
      "type": "http",
      "url": "https://api.yeelight.com/apis/metadata_mcp_server/v1/mcp",
      "headers": {
        "Authorization": "Bearer <YOUR_TOKEN>"
      }
    }
  }
}

README

Yeelight Metadata MCP Server

简体中文

Official Repository And Mirrors

GitHub is the canonical source for issues, contributions, CI, and releases. Read-only mirrors are available on Gitee and GitCode for users who cannot reach GitHub reliably, with GitLab.com as an additional global fallback. Clone from any reachable source, but report issues and contribute changes on GitHub.

This repository contains the home-understanding and management service inside Yeelight MCP. Ordinary users connect the complete Yeelight MCP suite once; they do not choose or install this server separately. Its guarded task interface covers homes, rooms, devices, groups, panels, scenes, automations, favorites, maintenance, and account operations.

[!TIP] New to Yeelight MCP? Follow the official English Yeelight MCP tutorial to prepare Yeelight Home, connect the complete Metadata and IoT capability suite, and verify the first real response. A Simplified Chinese tutorial is also available.

Yeelight Cloud MCP Suite

Yeelight Metadata MCP and Yeelight IoT MCP form the official Yeelight MCP cloud suite. They are independently deployed services configured together by the ordinary-user setup:

ServerRole in the suiteCore capabilities
Metadata MCPHome understanding and guarded management.Broad discovery and guarded workflows for homes, rooms, devices, groups, panels, scenes, automations, favorites, maintenance, accounts, multi-region authorization, and request-scoped home selection.
IoT MCPLive state and focused control.Direct topology and live-state access, device control through control_node, and scene execution through execute_scene.

Recommended composition: run one Yeelight MCP setup. It configures both services in the same AI client so the model receives home understanding, management, live state, and control without a second install or scan.

Place In Yeelight AI

yeelight-home provides the recommended installation, QR sign-in, and client-configuration flow. Ordinary users should prefer the complete yeelight-smart-home Skill. When a client cannot install Skills, one cloud setup connects the complete Yeelight MCP suite; the cloud services then execute independently of the local Runtime.

In plain language: this server gives an MCP-only AI a structured view of your home and guarded ways to manage it. It is not another CLI and it is not required by yeelight-home or the Smart Home Skill.

Choose this whenChoose something else when
Your AI client supports MCP but cannot install Agent Skills.Your client can install Skills: use yeelight-smart-home for the complete guided experience.
The AI needs rooms, devices, groups, scenes, automations, panels, or home management.You are scripting or troubleshooting in a terminal: use yeelight-home directly.
You need to develop or troubleshoot the home-understanding service inside Yeelight MCP.You only need an existing specialized control_node or execute_scene integration: use the IoT service documentation.

Highlights

  • Six namespaced tools instead of a large set of low-level HTTP operations.
  • 7 task groups, 23 tasks, and a runtime action catalog.
  • Safe-by-default execution: dryRun defaults to true.
  • Explicit confirmation for S2/S3 side effects.
  • Candidate actions are blocked from live execution by default.
  • MCP resources and prompts for task discovery and change planning.
  • Compatible METADATA_MCP_* and legacy APP_MCP_* environment variables.

Hosted Service

The official Streamable HTTP endpoint is:

text
https://api.yeelight.com/apis/metadata_mcp_server/v1/mcp

Required request header:

  • Authorization

Recommended optional headers:

  • Yeelight-Region: cn, sg, us, or eu; defaults to the deployment Region (cn by default)
  • House-Id

Authorization is the only required user credential. When Yeelight-Region is absent, the server can use the JWT Region claim to select an official endpoint, then validates the complete token there. Yeelight services derive client identity from that verified authorization context; users never enter a Client ID. For actions that need a home, precedence is context.houseId -> House-Id -> first Pro home in the same Region. The fallback is request-scoped, not a global current-home state.

MCP Capabilities

Tools:

  • yeelight_metadata.list_groups
  • yeelight_metadata.list_houses
  • yeelight_metadata.list_tasks
  • yeelight_metadata.list_actions
  • yeelight_metadata.get_action_schema
  • yeelight_metadata.execute_task

Resources:

  • yeelight-metadata://tasks
  • yeelight-metadata://actions
  • yeelight-metadata://schema/{task}/{action}

Prompts:

  • yeelight_metadata.inspect_home_metadata
  • yeelight_metadata.plan_metadata_change
  • yeelight_metadata.troubleshoot_automation_metadata

Treat the live MCP tools/list response as the source of truth for tool schemas.

Quick Start

Recommended: let Yeelight Home complete sign-in and configuration

bash
npm install --global yeelight-homeyeelight-home setup --lang en-US --mode mcp --agent cursor --mcp-source cloud --yes

The command displays the QR code. In Yeelight Pro app, tap Home's top-right +, choose MCP Authorization, and scan it. A single Pro home is selected automatically; only multiple homes require a choice by name. Replace cursor with your client or use --agent auto for detection. The generated client configuration starts a local credential proxy and contains no Authorization value. Manual token configuration remains an advanced compatibility path; never paste a token into an AI chat.

Connect to the hosted server

json
{  "mcpServers": {    "yeelight-metadata": {      "url": "https://api.yeelight.com/apis/metadata_mcp_server/v1/mcp",      "headers": {        "Authorization": "<YOUR_AUTHORIZATION>",        "Yeelight-Region": "cn",        "House-Id": "<YOUR_HOUSE_ID>"      }    }  }}

Both optional headers may be omitted. A JWT token supplies its Region claim; opaque tokens use the deployment default. Match an explicitly selected endpoint host to the Region (api.yeelight.com, api-sg.yeelight.com, api-us.yeelight.com, or api-de.yeelight.com). To switch homes, call yeelight_metadata.list_houses, then pass the chosen ID as request.context.houseId on subsequent calls.

See the integration guide for Cursor, Claude Desktop, local source, and combined IoT/Metadata configurations.

Run from source

Requirements: Python 3.10 or newer. Python 3.12 is recommended for local development.

bash
uv venv .venvsource .venv/bin/activateuv pip install -e '.[test]'
METADATA_MCP_RUNTIME_ENV=local ./service.sh start./service.sh status

The local endpoint is http://127.0.0.1:9010/mcp. A request without an Authorization header returns 401 by design.

Direct startup is also supported:

bash
METADATA_MCP_RUNTIME_ENV=local \METADATA_MCP_BIND_HOST=127.0.0.1 \PYTHONPATH=src \.venv/bin/uvicorn main:streamable_http_app --host 127.0.0.1 --port 9010

Safety Model

Every action has an execution mode, side-effect level, status, and parameter schema.

LevelMeaningExamples
S0Read-onlylist, get, search
S1Diagnostic or external activitydiagnose, download, poll
S2Create or updatecreate, update, configure
S3Destructive or identity-sensitivedelete, unbind, transfer, login

execute_task defaults to a plan-only request:

json
{  "request": {    "task": "family_space.manage_room",    "action": "create",    "payload": {"name": "Sample Room"},    "options": {"dryRun": true}  }}

Live S2/S3 execution requires confirmSideEffect=true. Candidate actions also require allowCandidate=true and are restricted by runtime policy. Start with a dry run, review plan.httpRequest, then execute only with testable or recoverable data.

Configuration

VariableDefaultPurpose
METADATA_MCP_RUNTIME_ENVprodprod, dev, local, or test
METADATA_MCP_BIND_HOSTlocal/test: 127.0.0.1; prod/dev: 0.0.0.0Listen address
METADATA_MCP_PORT9010Listen port
METADATA_MCP_PATH/mcpMCP route
METADATA_MCP_DEFAULT_REGIONcnDefault Region when Yeelight-Region is absent
METADATA_MCP_HOST_PREFIXhttps://api.yeelight.comYeelight API host
METADATA_MCP_API_BASE_URLhost prefixExplicit API base URL
METADATA_MCP_HTTP_TIMEOUT15Upstream timeout in seconds
METADATA_MCP_RESOURCE_DIRresources/catalogRuntime catalog override
METADATA_MCP_NACOS_ENABLEenvironment-dependentNacos registration switch
METADATA_MCP_NACOS_SERVERenvironment-dependentNacos servers
METADATA_MCP_ALLOW_CANDIDATE_EXECUTIONfalseCandidate execution gate

Legacy APP_MCP_* names remain supported for deployment compatibility. Standard Yeelight cloud origins support per-request Region routing. A custom API base is fixed to METADATA_MCP_DEFAULT_REGION; requests naming another Region are rejected instead of being silently misrouted.

Documentation

Development

bash
PYTHONPATH=src python -m pytest -qpython -m compileall -q main.py src tests scripts

The repository also includes protocol and acceptance utilities under scripts/. Live acceptance commands can change cloud data; use explicit credentials, a non-production home, and recoverable fixtures.

License

Licensed under the Apache License 2.0.

來源:README.md,提交 b34ca72

工具

0
該伺服器需要 API 金鑰等憑證,因此 SourceWeft 無法在沒有憑證的情況下讀取其工具,也未收錄。

版本歷史

1
  1. v1.0.0最新Oct 12, 2026

評分

已滿足適用檢查項的 85%

該評分衡量此收錄條目的完整程度,依據註冊中心條目、端點與公開來源計算得出。它不是對伺服器本身的評價,也不表示伺服器安全或目前可用:可連通性是帶日期的檢查結果,評分也不包含安裝量或使用者評價。

  • 近 30 天內可連通必需25 分
  • 已宣告認證方式必需10 分
  • 已列出工具15 分

    請讓伺服器在無憑證時回應 tools/list,或發布可在沙箱中無憑證啟動的軟體包。

  • 描述必需15 分
  • 圖示10 分
  • 文件15 分
  • 授權條款10 分