Eight Sleep MCP
Unofficial MCP server for Eight Sleep, reverse-engineered from observed mobile app traffic and validated against live device behavior.
README
Eight Sleep MCP
Unofficial MCP server for Eight Sleep, reverse-engineered from observed mobile app traffic and validated against live device behavior.
Built entirely by OpenAI GPT-5.4 via Codex.
Overview
This project exposes Eight Sleep account, device, temperature, and sleep data through MCP, and supports the core temperature controls confirmed from live app traffic.
Supported capabilities include:
- reading account, device, temperature, recent temperature events, and sleep intervals
- reading a compact summary of the most recent sleep interval without timeseries payloads
- fetching compact interval pages with
get_sleep_intervalsview: "summary"when full timeseries are not needed - resuming smart temperature mode
- turning temperature control off
- applying a temporary bedtime temperature change
- saving a bedtime temperature change to the active schedule
- updating saved smart temperature levels
- toggling Autopilot
Behavior Model
- Temporary bedtime changes are sent as override values to the pod temperature endpoint.
- Permanent bedtime changes update the bedtime schedule, then clear the temporary override.
- Smart temperature profile editing is separate from bedtime schedule editing.
- Turning temperature off disables active heating and cooling; it does not power down the hardware.
API Coverage
The current implementation uses confirmed requests across auth-api, client-api, and app-api, including authentication, user and device reads, temperature state, temperature events, bedtime schedule updates, pod temperature control, and Autopilot mode changes.
Sleep Data Guidance
For LLM-facing clients, use the compact sleep tools by default.
Golden paths:
- Use
get_latest_sleep_summaryfor the most recent night when you only need compact summary metrics. - Use
get_sleep_intervalswithview: "summary"when you need multiple recent intervals but do not needtimeseries,stages, orsnoring.
Example get_latest_sleep_summary response:
{
"sleepStart": "2026-04-02T03:38:30.000Z",
"sleepEnd": "2026-04-02T15:35:30.000Z",
"sleepDuration": 34380,
"deepPct": 0.12,
"remPct": 0.24,
"lightPct": 0.64
}
Example compact interval call:
{
"tool": "get_sleep_intervals",
"arguments": {
"pages": 3,
"view": "summary"
}
}
sleepStart and sleepEnd are returned as UTC ISO timestamps. Convert them to the user's local timezone in the caller if you need local bedtime or wake time.
Setup
npm install
cp .env.example .env
npm test
npm run build
Configuration supports either:
EIGHT_SLEEP_ACCESS_TOKEN- or
EIGHT_SLEEP_EMAIL,EIGHT_SLEEP_PASSWORD,EIGHT_SLEEP_CLIENT_ID, andEIGHT_SLEEP_CLIENT_SECRET
Optional values:
EIGHT_SLEEP_USER_IDEIGHT_SLEEP_CLIENT_DEVICE_IDEIGHT_SLEEP_REQUEST_TIMEOUT_MS
MCP Bundle
For bundle-aware clients, a manifest.json is included and a local MCP bundle can be built with:
npm run package:mcpb
That produces an .mcpb file for one-click installation in clients that support MCP bundles.
MCP Setup
The server runs over stdio.
Example client configuration for the built server:
{
"mcpServers": {
"eight-sleep": {
"command": "node",
"args": ["/absolute/path/to/eight-sleep-mcp/dist/index.js"],
"cwd": "/absolute/path/to/eight-sleep-mcp"
}
}
}
For local development without building first:
{
"mcpServers": {
"eight-sleep": {
"command": "npx",
"args": ["tsx", "src/index.ts"],
"cwd": "/absolute/path/to/eight-sleep-mcp"
}
}
}
If a client does not support cwd, pass the Eight Sleep environment variables directly in the client configuration instead of relying on .env.
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。