HealthCoach: AI Health Coach

Problem

Many people want to manage exercise, but common tools either depend on hardware or only log data without generating plans based on current status or adjusting over time. HealthCoach 2.0 aims to solve this: without a full health platform, how to turn exercise check-ins, exported device data, heart-rate readings, and activity metrics into an actionable daily plan, with reminders and summaries forming a feedback loop.

How It Works

The skill is built around a PDCA loop: plan → execute → evaluate → adjust. It supports two data collection paths:

• Hardware sync: currently supports COROS, Apple Watch, Xiaomi bands/watches, and Huawei watches/bands; Garmin is listed as planned.
• Manual check-ins: you can say “I exercised today” or “I want to check in,” and the skill guides the recording process.

For analysis, the materials mention multi-agent reasoning across Western medicine, traditional Chinese medicine, exercise science, and nutrition to generate a dynamic, personalized exercise plan. Fixed reminders include 09:00 daily plan, 18:00 check-in reminder, and 21:00 daily summary plus next-day plan. Data is stored locally at ~/Downloads/HealthSkill2.0/ and is not uploaded to the cloud.

Limits

It is not a medical device and cannot replace professional diagnosis or treatment advice. Manual mode depends on accurate user input; hardware coverage is still expanding, and Garmin is not yet available. It is well suited for a personal health assistant focused on exercise logging, plan generation, and reminder loops, but not for use as the sole source of health decisions.