Athlete
Evidence-Based Fitness Programming — From Screenshots to Training Plan
Key Takeaway
Send your Garmin, Apple Watch, Whoop, or Strava screenshots to your AI agent. It extracts every data point, analyzes your training load, recovery status, and performance trends, then builds a progressive 8-12 week plan with built-in deload triggers and safety guardrails.
The Problem
You have a fitness tracker generating beautiful data. You look at the graphs. You think "huh, my HRV is down." Then you do the same workout you were going to do anyway because you don't know what the data means for your training.
The gap between data collection and data-driven training is enormous:
- Garmin Connect shows you charts but doesn't program your workouts based on them
- Apple Fitness gives you rings but no periodization
- Whoop tells you your recovery score but doesn't adjust your plan
- Strava shows you segments but doesn't prevent overtraining
Hiring a coach who actually uses your wearable data costs $200-500/month. Most coaches still program by feel, not by data.
The Solution
The Fitness Coach skill reads your fitness screenshots like a human coach would — extracting every metric, contextualizing it against your baseline, and using it to build or adjust your training plan.
The Process
Data extraction from screenshots:
You send a Garmin Connect screenshot showing your last week of training:
View details
Agent:
## Data Extraction — Garmin Connect (Mar 7-13)
| Date | Activity | Duration | Distance | Avg HR | Max HR | TE |
|------|----------|----------|----------|--------|--------|-----|
| Mar 7 | Run | 45:12 | 7.2km | 148 | 172 | 3.8 |
| Mar 8 | Strength | 52:00 | — | 128 | 155 | 3.1 |
| Mar 9 | REST | — | — | — | — | — |
| Mar 10 | Run | 38:30 | 6.1km | 152 | 178 | 4.0 |
| Mar 11 | Run | 62:15 | 10.4km | 142 | 165 | 3.5 |
| Mar 12 | Strength | 48:00 | — | 125 | 150 | 2.9 |
| Mar 13 | Run | 32:00 | 5.3km | 155 | 180 | 4.2 |
Weekly Summary:
- Running: 4 sessions, 29km total, avg pace 6:15/km
- Strength: 2 sessions, 100 min total
- Training Load: 680 (your 7-day average)
- VO2 Max estimate: 44
- Resting HR trend: 56 → 59 BPM (rising ⚠️)
- HRV: 42ms (down from 52ms baseline ⚠️)
- Sleep: 6.4h average (below 7h target ⚠️)
## ASSESSMENT
⚠️ OVERREACHING SIGNALS DETECTED:
1. Resting HR rising (+3 BPM over 7 days)
2. HRV declining (-19% from baseline)
3. Sleep consistently under target
4. Training effect scores rising (working harder
for same output = fatigue accumulation)
Your Training Load of 680 is at the upper end of
your productive range (450-700). You're not
overtrained, but you're knocking on the door.
## RECOMMENDATION: Modified Week
This week should be a DELOAD:
- Reduce running volume by 40% (29km → ~17km)
- Keep 2 strength sessions but reduce weight by 20%
- Prioritize sleep (add 30 min — this is non-negotiable)
- One complete rest day mid-week
- NO high-intensity sessions until HRV recovers to 48+
## NEXT 8-WEEK BLOCK (starting after deload):
Week 1-2: Base building (4 runs, 30-35km/week, all easy)
Week 3-4: Introduce 1 tempo run per week (+5km volume)
Week 5-6: Add 1 interval session, drop 1 easy run
Week 7: Peak week (~42km, 1 tempo + 1 interval)
Week 8: Taper / deload
Progression trigger: Advance to next phase when
resting HR stable AND HRV within 10% of baseline
for 5+ consecutive days.
Safety trigger: If resting HR rises >5 BPM above
baseline for 3+ consecutive days → mandatory deload
regardless of schedule.
The Results
| Metric | Self-Programmed | App (Garmin Coach) | AI Fitness Coach |
|---|---|---|---|
| Uses YOUR data | Partially | Partially | Fully (all metrics) |
| Overtraining detection | No (ignoring signals) | Basic | Multi-metric correlation |
| Plan adaptation | None (same plan regardless) | Weekly | Real-time |
| Injury prevention | Low | Medium | High (safety triggers) |
| Cost | Free (but risky) | Free | ~$0.10/week |
| Periodization | Random | Template | Personalized to data |
Setup on MrChief
yamlShow code
skills:
- fitness-coach
- apple-health-skill # or strava-skill
Related case studies
Patient
Holistic Wellness Dashboard — Sleep, Stress, Nutrition, Movement in One View
Combine the wellness skill with sleep, nutrition, and fitness data to get a single holistic health assessment. No more checking 5 different apps to understand how you're actually doing.
Athlete
Strava Activity Intelligence — Your Agent Knows When You Should Rest
The Strava Skill syncs your running, cycling, and swimming data to your AI agent. It analyzes weekly mileage trends, detects overtraining patterns, compares performance across periods, and tells you when to push and when to rest — based on YOUR data, not generic advice.
Biohacker
Testosterone Optimization Dashboard — Track What Actually Moves the Needle
The Testosterone Optimization skill creates a personal tracking system for the 6 evidence-based factors that influence natural testosterone production. After 4 weeks of data, it reveals which specific lever — sleep, lifting, vitamin D, zinc, stress, or body fat — makes the biggest difference for YOUR biology.
Want results like these?
Start free with your own AI team. No credit card required.