Oura MCP server

Oura MCP server

Enables users to access and query Oura ring health data, including sleep periods, readiness, activity, workouts, heart rate, stress, SpO2, resilience, and more, through Claude.ai or Claude Code using the Oura API v2 with OAuth 2.1 login.

Category
访问服务器

README

<center align="center" style="text-align: center;justify-content:center;"> <div align="center" style="text-align: center;justify-content:center;"> <h1 align="center" style="text-align: center;justify-content:center;">

Oura MCP server

<img style="justify-content:center;text-align: center;width: 95px; height: auto;" width="793" height="411" alt="image" src="https://github.com/user-attachments/assets/abed1a04-d69b-4ab4-a490-d606064df72d" /> <img style="justify-content:center;text-align: center;width: 145px; height: auto;" alt="image" src="public/oura-wordmark-white.png" />

</h1>

Version Python Node OAuth

</div> </center>

<hr>

Read your Oura ring from Claude.ai and Claude Code: sleep with the full periods behind the score, readiness, activity, workouts, heart rate, stress, SpO2, resilience, cardiovascular age, VO2 max, bedtime recommendations, tags and rest mode. It talks to the documented Oura API v2 and puts an OAuth 2.1 login in front so you can add it to Claude.ai as a custom connector. Claude Code can use a plain token instead.

<hr>

Why not the npm package

This started as a wrapper around @daveremy/oura-mcp and replaced it, because that package returns nothing for most of what you would ask it. It queries single-day ranges as start_date == end_date, and the Oura API does not answer those the way the naming suggests:

  • sleep filters on bedtime_start in UTC, so a night that begins at 00:31 local (UTC+3) lands on the previous UTC day and falls outside its own day's window
  • daily_activity and workout treat end_date as exclusive, making a same-day range zero-width

The result is silent: oura_sleep returns a score with an empty periods array, oura_activity returns null, and oura_workouts returns [], on every single-day call. Nothing errors, so it reads as "no data recorded" rather than "the query was wrong". This server pads the range on both sides and filters back by each record's own day field, which is timezone-correct, and follows next_token to exhaustion instead of silently returning one page.

Tools

Tool What you get
oura_daily_summary Sleep score, readiness score, sleep periods and activity for a day, in one call
oura_sleep The daily score with its contributors, plus the raw periods: duration, stages, heart rate, HRV
oura_sleep_time Oura's recommended bedtime window, and how the actual bedtime compared
oura_readiness Readiness score and its contributors
oura_resilience Long-term resilience level, with sleep and daytime recovery contributors
oura_cardiovascular_age Estimated vascular age and pulse wave velocity
oura_vo2_max Estimated VO2 max
oura_activity Score, steps, calories, distance and the activity time breakdown
oura_workouts Workouts with heart rate, calories, duration and intensity
oura_heart_rate Continuous heart rate samples for a time window
oura_stress Daily stress and recovery time
oura_spo2 Average nightly blood oxygen
oura_sessions Meditation, breathing and relaxation sessions
oura_tags Tags logged for a day, both the legacy and the enhanced kind
oura_rest_mode Rest mode periods covering a day
oura_ring Every ring ever paired, plus battery readings
oura_personal_info Age, weight, height, biological sex, email
oura_trends Sleep, readiness, activity and stress scores over a range of days

Every date-scoped tool takes an optional date as YYYY-MM-DD and defaults to today. Oura files a night under the day you woke up.

oura_vo2_max, oura_sleep_time and oura_cardiovascular_age are sparse by design: Oura computes them periodically rather than daily, so an empty result is normal and not a sync failure.

How it fits together

Claude.ai / Claude Code
        |  HTTPS
   Cloudflare Tunnel, or any proxy that gives you HTTPS
        |
   nginx  127.0.0.1:8441
        |
   auth-server.js  :8442    handles the login and the tokens
        |
   oura-mcp  :8440          streamable HTTP, local only
        |
   api.ouraring.com

Ports are 8440-8442 so they sit clear of the other MCPs on this host. The server on :8440 has no login of its own and refuses to bind anywhere but loopback, so it is only ever reached through the login layer.

Setup

Needs Python 3.12 or newer, and Node for the login layer.

python3 -m venv .venv
.venv/bin/pip install -e .
npm install --omit=dev

mkdir -p ~/.config/oura-mcp && chmod 700 ~/.config/oura-mcp
echo "OURA_TOKEN=your-personal-access-token" > ~/.config/oura-mcp/env
chmod 600 ~/.config/oura-mcp/env

CONFIG_DIR=~/.config/oura-mcp node set-password.js 'your-password-here'
openssl rand -hex 32 > ~/.config/oura-mcp/token
chmod 600 ~/.config/oura-mcp/token

Get the Oura token from cloud.ouraring.com.

Then fill in YOUR_USER and the hostname in systemd/*.service and nginx/oura-mcp.conf, and:

sudo cp systemd/*.service /etc/systemd/system/
sudo systemctl daemon-reload
sudo systemctl enable --now oura-mcp oura-mcp-auth
sudo cp nginx/oura-mcp.conf /etc/nginx/sites-enabled/oura-mcp
sudo nginx -t && sudo systemctl reload nginx

Point a tunnel or an HTTPS proxy at 127.0.0.1:8441. OAuth needs HTTPS.

Check it from outside: discovery should return metadata, and /mcp without a token must return 401.

curl https://your-host/.well-known/oauth-authorization-server
curl -o /dev/null -w '%{http_code}\n' -X POST https://your-host/mcp

Connecting

Claude.ai: Settings, Connectors, Add custom connector, https://your-host/mcp, client ID and secret blank.

Claude Code:

claude mcp add --transport http oura https://your-host/mcp \
  --header "Authorization: Bearer $(cat ~/.config/oura-mcp/token)" --scope user

Running it over stdio instead, without any of the login layer:

OURA_TOKEN=your-token .venv/bin/oura-mcp

Credits

The login layer comes from rollecode/obsidian-remote-mcp.

推荐服务器

Baidu Map

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。

官方
精选
JavaScript
Playwright MCP Server

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。

官方
精选
TypeScript
Magic Component Platform (MCP)

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。

官方
精选
本地
TypeScript
Audiense Insights MCP Server

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。

官方
精选
本地
TypeScript
VeyraX

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。

官方
精选
本地
graphlit-mcp-server

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。

官方
精选
TypeScript
Kagi MCP Server

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。

官方
精选
Python
e2b-mcp-server

e2b-mcp-server

使用 MCP 通过 e2b 运行代码。

官方
精选
Neon MCP Server

Neon MCP Server

用于与 Neon 管理 API 和数据库交互的 MCP 服务器

官方
精选
Exa MCP Server

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。

官方
精选