airbone42

airbone42 / 360-data-athlete

Public

This is a coding experiment with multi-agent systems, using sport training as the problem domain. It does not replace a coach or sports-medical advice. Use only with a solid training background, at your own risk. No warranty, no support, no audit.

16
2
85% credibility
Found May 24, 2026 at 17 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
Python
AI Summary

360° Data Athlete is an AI coaching framework that runs as a plugin inside Claude Code. It pulls your fitness data from intervals.icu, Garmin, and Strava, then uses a team of specialized AI agents to plan your daily training—deciding what to do, designing the specific workouts, validating them for safety, and pushing them to your fitness tracker. After you train, it analyzes your performance and gives you coaching feedback. The system tracks muscle fatigue, monitors your HRV response patterns, and audits itself for consistency. It is explicitly experimental, not a replacement for human coaching, and requires an intervals.icu account plus Python 3.11+ to run.

How It Works

1
📱 You hear about an AI coach for athletes

A friend tells you about a system that uses AI to plan your training based on your fitness data and daily readiness.

2
⚙️ You connect your fitness accounts

You link your intervals.icu account so the system can see your heart rate variability, training load, and past workouts. Optionally connect Strava for gear tracking and Garmin for running dynamics.

3
🌅 Each morning, your status is checked

Before you even ask, the system pulls your latest wellness data—HRV, resting heart rate, sleep quality—and checks the weather and your upcoming race calendar.

4
🧠 An AI planning team builds your day

A head planner decides what to train, then specialist agents design the specific workouts—one for running, one for strength, one for ninja athletics—each with your history and restrictions in mind.

5
✅ You review and approve the plan

You see a clear, readable plan with shoe recommendations and reasoning. You say 'go' or push back if something feels off, and the system adjusts.

6
📲 Workouts land in your fitness app

With one click, your planned sessions appear in intervals.icu, ready for your GPS watch to sync. After you train, Strava gets updated with your activity names.

7
📊 You get coaching feedback after each session

The system analyzes your laps, compares your HRV response to expected patterns, and gives you 2-3 concrete takeaways—no generic praise, just specific observations.

🏆 Your training stays consistent and balanced

Over weeks, the system tracks muscle fatigue across your whole body, warns when you're overdue for certain exercises, and keeps your weekly hard sessions in balance—all while respecting when you need rest.

Sign up to see the full architecture

6 more

Sign Up Free

Star Growth

See how this repo grew from 17 to 16 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is 360-data-athlete?

This is an AI-powered training coach that runs as a Claude Code plugin, using a team of specialized AI agents to plan, push, and review endurance and strength training. It pulls data from intervals.icu, Strava, and Garmin to build a daily training context—HRV, fitness metrics, sleep, weather, race calendar—and generates personalized workout plans with pre-push validation and post-activity analysis. Built in Python, it uses plain-text config files and markdown prompts rather than a traditional framework, making the agent logic transparent and editable.

Why is it gaining traction?

The architecture is the hook: instead of building another LangChain wrapper, this project asks "what happens when you treat Claude Code as a general-purpose agent harness and point it at a real-world domain?" The answer is a multi-agent system where each sub-agent (planner, workout specialists, mental coach, plan validator) runs in isolated context with cross-workout consistency checks. The HRV forecasting model that predicts expected heart-rate response after training sessions is genuinely novel. The maintainer is a competitive athlete who built this for their own training, which means the sports-science logic is grounded in actual practice, not generated fluff.

Who should use this?

This is for experienced endurance athletes who want to understand how AI reasoning applies to their training—not for beginners or anyone without a solid training background. You need to be comfortable editing markdown configs, running Python scripts, and reading along with how the system makes decisions. If you're a developer interested in Claude Code as an agent harness or want to explore multi-agent orchestration outside of code generation, this is a legitimate test bench. If you want a polished product with support and stability guarantees, look elsewhere.

Verdict

With a 0.85% credibility score and only 16 stars, this is an early-stage experiment by a solo maintainer—no commercial backing, no formal support structure. The documentation is thorough and the architecture is thoughtful, but it's explicitly unstable and evolves weekly. For developers exploring AI coaching or Claude Code agent patterns, it's worth the investment. For athletes seeking reliable, production-ready training software, this is not ready yet.

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.