Async

Context

Health and fitness information was spread across individual readings and app views, making it difficult to compare activity, workout, heart-rate, and recovery patterns over time.

Problem

Raw metrics can be overwhelming and easy to misinterpret. I wanted a lightweight reflection tool that preserved prior context without presenting uncertain patterns as diagnoses.

What I built

I connected Apple Watch data through a health-data MCP to a Codex-powered agent and created scheduled workflows that review recent readings against earlier periods, identify meaningful changes, and preserve context for follow-up.

How it works

1. 01Retrieve available Apple Watch health and fitness metrics through the connected MCP.
2. 02Compare recent activity, workout, heart-rate, recovery, and related readings with previous periods.
3. 03Identify meaningful changes or unusual patterns without converting them into medical conclusions.
4. 04Generate a concise, structured update and schedule future reviews through the Codex agent.
5. 05Preserve previous observations so later updates can be evaluated longitudinally.

Outcome

• Reduced the need to manually compare readings across multiple health views.
• Created a persistent, scheduled record of personal fitness and recovery patterns.
• Explored safety-aware communication for sensitive, uncertain health data.

What I learned

Context-aware health tracking is useful when it helps a person notice patterns without overstating certainty. The product boundary between reflection and diagnosis must remain explicit.