oura

This MCP server provides programmatic access to the Oura API, exposing tools to fetch sleep, readiness, and resilience data. It supports both date-range queries and today’s data, enabling language models to retrieve structured metrics for specific periods or for the current day. To use it, supply your Oura API token as an environment variable so the server can authenticate requests to the Oura API. The available tools include get_sleep_data(start_date, end_date), get_readiness_data(start_date, end_date), get_resilience_data(start_date, end_date), and their today variants: get_today_sleep_data(), get_today_readiness_data(), get_today_resilience_data(). Dates should be provided in ISO format (YYYY-MM-DD).

In Claude/Claude Desktop, you configure the MCP server in claude_desktop_config.json by referencing the uvx command and the package name so Claude can route queries to the server. For example, the config maps the server name "oura" to the uvx command that runs the oura-mcp-server package, with the OURA_API_TOKEN supplied as an environment variable.

Prerequisites:

• Python 3.12 (as indicated by the project), and pip
• An Oura API Personal Access Token

Installation steps:

1. Ensure Python and pip are up to date

• On macOS/Linux: python3 -m pip install --upgrade pip
• On Windows: py -m pip install --upgrade pip

2. Install the Oura MCP Server package

• Option A (pip install): pip install oura-mcp-server
• Option B (via uvx usage with a local install): pip install --user oura-mcp-server

3. Set the required environment variable for authentication

• Export the token (Unix/macOS): export OURA_API_TOKEN=YOUR_OURA_API_TOKEN
• Set the token (Windows PowerShell): $env:OURA_API_TOKEN = "YOUR_OURA_API_TOKEN"

4. Run the MCP server (via uvx as shown in the README)

• uvx oura-mcp-server

Optional alternatives:

• Install with pipx (if you prefer isolated environments): pipx install oura-mcp-server export OURA_API_TOKEN=YOUR_OURA_API_TOKEN uvx oura-mcp-server

Notes:

• The server requires the OURA_API_TOKEN to access user data from the Oura API.
• If you’re deploying in an environment without uvx, you can also run in a Python environment directly by invoking the package's entry point if available (refer to the package's CLI documentation).

Tips and considerations:

• Ensure the OURA_API_TOKEN you provide has the necessary scopes for the data you intend to access (sleep, readiness, resilience).
• If you encounter authentication errors, double-check token validity and that it hasn’t expired.
• For Claude integration, ensure claude_desktop_config.json points to the correct command and package name, and that the environment variable OURA_API_TOKEN is set in that environment.
• ISO date formatting is strict (YYYY-MM-DD); validate inputs to avoid date parsing errors.
• If the MCP server cannot reach the Oura API due to network restrictions, verify outbound connectivity from the host running the server.