oura-mcp-server
Enables AI assistants to query Oura Ring health data (sleep, readiness, activity, heart rate, SpO2, stress) via a remote MCP server.
README
Oura MCP Server
A small remote MCP server that lets Claude and ChatGPT pull your Oura Ring data (sleep, readiness, activity, heart rate, SpO2, stress) live, on demand. Your ring already syncs to Oura's cloud through the phone app — this reads from that cloud API, so there's nothing to install on your phone.
Tools exposed
| Tool | What it returns |
|---|---|
oura_get_daily_summary |
Sleep + readiness + activity + SpO2 + stress for a date range, in one call |
oura_get_daily_sleep |
Nightly sleep score + contributors |
oura_get_daily_readiness |
Readiness score + recovery contributors (best recovery signal) |
oura_get_daily_activity |
Activity score, steps, calories, MET minutes |
oura_get_sleep_periods |
Detailed sessions: stages, HR, HRV, breathing |
oura_get_heart_rate |
Time-series heart-rate samples |
oura_get_daily_spo2 |
Blood oxygen % during sleep |
oura_get_daily_stress |
Daytime stress + recovery time |
oura_get_personal_info |
Profile (age, sex, height, weight) |
All accept start_date / end_date as today, yesterday, or YYYY-MM-DD.
Setup (about 15 minutes, once)
1. Get Oura OAuth credentials
Oura phased out Personal Access Tokens (Dec 2025), so use OAuth2 — it also lets the server refresh itself so daily pulls keep working unattended.
- Go to https://cloud.ouraring.com/oauth/applications and create an application.
- Set the Redirect URI to exactly
http://localhost:8080/callback. - Copy the Client ID and Client Secret.
2. Get your refresh token (one time)
On your computer (Node 20+):
npm install
OURA_CLIENT_ID=xxx OURA_CLIENT_SECRET=yyy node get-token.mjs
Approve in the browser window that opens. The terminal prints OURA_REFRESH_TOKEN=.... Save it.
Just testing? Skip steps 1–2 and set
OURA_USE_SANDBOX=trueto use Oura's sample data.
3. Pick an MCP secret
Generate a long random string (e.g. openssl rand -hex 24). This protects your endpoint and becomes part of the URL.
4. Deploy (always-on cloud)
Push this folder to a GitHub repo, then use any host. Set these environment variables on the host: MCP_SECRET, OURA_CLIENT_ID, OURA_CLIENT_SECRET, OURA_REFRESH_TOKEN.
- Railway (no cold starts): New Project → Deploy from repo → add the env vars. Done.
- Render: New → Web Service → connect repo (a
render.yamlis included) → add env vars. - Docker (any host / Fly.io): a
Dockerfileis included.
Your connector URL is:
https://<your-deployed-host>/mcp/<MCP_SECRET>
Confirm it's live: visiting https://<your-deployed-host>/health should return {"status":"healthy"}.
Connect the AI apps
Claude (Pro, Max, Team, or Enterprise)
Settings → Connectors → Add custom connector → paste the URL above → save. Start a chat and ask: "What was my Oura readiness this week?"
ChatGPT (Plus, Pro, Business, Enterprise, or Edu)
Settings → Connectors / Apps → enable Developer mode → Add custom connector → paste the URL → save. (ChatGPT only supports remote HTTPS MCP servers, which this is.)
Both clients connect from the cloud, so the server must be publicly reachable over HTTPS (any of the hosts above provide that automatically).
Daily use — example prompts
- "Give me my Oura daily summary for today."
- "How has my sleep score trended over the last 14 days?"
- "Compare my readiness on days I hit my activity target vs days I didn't."
- "What was my average resting heart rate and HRV this week?"
Local run (for testing)
npm install && npm run build
MCP_SECRET=test OURA_USE_SANDBOX=true OURA_CLIENT_ID=x OURA_CLIENT_SECRET=x OURA_REFRESH_TOKEN=x npm start
# POST http://localhost:3000/mcp/test
Notes
- Read-only: the server never writes to your Oura account.
- If pulls start failing with a 401, your refresh token was likely revoked — re-run
get-token.mjsand updateOURA_REFRESH_TOKEN. - Keep
MCP_SECRETprivate; anyone with the full URL can read your data.
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。