Skip to content

Connect AI to your Aidlab data ​

Aidlab MCP lets a compatible AI client read your recorded data after you sign in and approve access: steps, distance, calories, workouts, heart rate, resting heart rate, HRV, respiration rate, SpO2, skin temperature, stress, weight, VO2max and sleep.

Connect your account ​

  1. In your AI client's MCP settings, add a remote server using Streamable HTTP and this URL:

    text
    https://my.aidlab.com/mcp
  2. Start the client's authorization flow. Sign in to Aidlab and check which app is asking before you grant access.

  3. Go back to your AI client and ask about your data.

Your client needs to support MCP authorization (OAuth). It registers itself automatically, so you don't create an API client or paste a token. Regular Aidlab API tokens don't work with this endpoint.

Try a question ​

How did I sleep this week? Compare it with my resting heart rate and HRV.

Show my heart rate yesterday, hour by hour, and mark my workouts.

How have my weight and daily steps changed over the last six months?

Answers depend on what your account has recorded. A day without a measurement is missing, not zero.

What the AI client can call ​

ToolWhat it returns
aidlab_get_daily_summaryOne row per day: activity, heart, respiration, SpO2, skin temperature, stress, weight and the night's sleep.
aidlab_list_measurementsOne metric as individual samples or hourly and daily values, up to a year back for daily values.
aidlab_list_sleepSleep stages.
aidlab_list_activitiesRecorded workouts.
aidlab_get_profileYour language and timezone, with the optional profile:read scope.

Days and hours follow the timezone the client asks for and always run from midnight to midnight. Sleep counts toward the day you woke up.

What you share ​

The data:read scope gives read-only access to your own recorded data. MCP tools can't change or delete it. The optional profile:read scope shares only your language and timezone.

The connected AI provider receives the data it requests and an access token to make those requests. Review that provider's data handling before you connect it.

Configuration examples and a smoke test are on GitHub. For integrations you build yourself, see OAuth and API access.