Leo Health 🫀
Your Apple Health and Whoop data — unified in a local SQLite database. In 60 seconds.
Apple Health locks your biometrics in a 4GB XML file. Whoop buries yours in CSVs with inconsistent column names. Leo Core parses both in under 60 seconds and writes everything to a single, normalized SQLite database — Heart Rate, Sleep, Workouts, HRV, Recovery Score, Blood Oxygen — all queryable with standard SQL.
Zero network requests. Runs locally. MIT licensed.
🧠 Leo Max — AI Health Coach (Coming Soon)
Local LLM that analyses your data privately. Bloodwork history, wearable trends, medical literature — all on your Mac, nothing leaves your machine.
Join the waitlist →
What it looks like
Install
git clone https://github.com/sandseb123/Leo-Health-Core.git
cd Leo-Health-Core
pip3 install -e .
Three commands are now available anywhere on your Mac:
leo # view your health dashboard
leo-watch # start watching Downloads for new exports
leo-dash # open full web dashboard in browser
Note: If pip3 is not found, try pip install -e . or install Python 3.9+ from python.org
Docker
Run Leo Health in a container — works on macOS, Linux, and Windows.
# Clone the repo
git clone https://github.com/sandseb123/Leo-Health-Core.git
cd Leo-Health-Core
# Start the dashboard
docker compose up -d
# Open in browser
open http://localhost:5380
Drop your health exports into the imports/ folder — Leo will detect and ingest them automatically.
# Import Apple Health export
cp ~/Downloads/export.zip imports/
# Import Whoop CSVs
cp ~/Downloads/whoop_recovery.csv imports/
Or run directly without compose:
docker run \
-p 127.0.0.1:5380:5380 \
-v ~/.leo-health:/data \
-e LEO_HOST=0.0.0.0 \
-e LEO_DB_PATH=/data/leo.db \
leo-health
Data is stored in a Docker volume (leo-data) and persists across container restarts.
# Stop
docker compose down
# View logs
docker compose logs -f
Get your data in
Apple Health (iPhone):
- Open the Health app
- Tap your profile picture → Export All Health Data
- AirDrop it to your Mac — Leo detects and parses it automatically
Whoop: Open Whoop app → Profile → Export Data → check email for CSVs → AirDrop to Mac
leo-watch # start the watcher — detects exports within 10 seconds
Auto-Ingest via AirDrop ✨
Leo watches your Downloads folder and automatically parses any health export the moment it arrives — no commands needed after setup.
- Checks for new files every 10 seconds
- Uses ~8MB RAM, near-zero CPU while idle
- Never processes the same file twice
- Sends a macOS notification when ingestion completes
Auto-start on login: Coming soon.
Query your data
sqlite3 ~/.leo-health/leo.db
-- Last 7 days of Whoop recovery
SELECT recorded_at, recovery_score, hrv_ms, resting_heart_rate
FROM whoop_recovery
ORDER BY recorded_at DESC LIMIT 7;
-- HRV by source (Apple Watch vs Whoop)
SELECT source, ROUND(AVG(value), 1) as avg_hrv, COUNT(*) as readings
FROM hrv GROUP BY source;
-- Sleep stage breakdown
SELECT stage, COUNT(*) as sessions
FROM sleep GROUP BY stage ORDER BY sessions DESC;
-- Top workouts by volume
SELECT activity, COUNT(*) as sessions, ROUND(AVG(calories), 0) as avg_cal
FROM workouts GROUP BY activity ORDER BY sessions DESC;
What gets parsed
Apple Health (export.zip)
| Data | Table | Metrics |
|---|
| Heart Rate | heart_rate | BPM, resting HR, walking avg |
| HRV | hrv | SDNN in milliseconds |
| Sleep | sleep | REM, Deep, Core, Awake stages |
| Workouts | workouts | Activity, duration, distance, calories |
| Blood Oxygen | blood_oxygen | SpO₂ % |
Whoop (CSV exports)
| Data | Table | Metrics |
|---|
| Recovery | whoop_recovery | Score, HRV, resting HR, SpO2 |
| Strain | whoop_strain | Day strain, calories, max/avg HR |
| Sleep | sleep | Performance %, time in bed, stages |
Privacy
Leo Core contains zero outbound network code. Verify the entire repository yourself:
grep -r "import urllib\|import http\|import requests\|import socket\|http.client\|urllib.request" .
You'll see three results, all in leo_health/dashboard.py — these are Python stdlib imports used to run a local web server on your own machine (localhost only). No data is sent anywhere. No outbound connections. No tracking.
Your data lives in ~/.leo-health/leo.db and never leaves your machine.
Threat model: Dashboard binds to localhost only. Anyone with
local machine access can read the DB. No authentication —
designed for single-user desktop use only.
Who this is for
Leo Core is an open-source developer tool. If you're comfortable with Terminal, git clone and bash install.sh gets you running in 2 minutes. Free forever, MIT licensed.
Leo Max is the upcoming AI health coach layer — a local LLM that runs against your unified health database, cross-references medical literature, and lets you upload bloodwork PDFs to track lab panels over time. Nothing leaves your Mac. Join the waitlist →
Project structure
leo_health/
├── parsers/
│ ├── apple_health.py # SAX streaming parser for export.zip
│ └── whoop.py # Auto-detecting CSV parser
├── db/
│ ├── schema.py # SQLite schema
│ └── ingest.py # Unified ingest for all sources
└── dashboard.py # Local web dashboard (localhost only)
status.py # leo command — terminal dashboard
watcher.py # leo-watch — auto-ingest on AirDrop
install.sh # One-command installer for macOS
pyproject.toml
Architecture Overview
flowchart TD
subgraph External Inputs
A[Apple Health Export XML]
B[Whoop CSV Export]
C[AirDrop Watch Folder]
end
subgraph Leo Health Core
D[Streaming SAX Parser]
E[CSV Parser]
F[Normalization & Validation]
G[(SQLite<br/>User-Owned DB)]
H[Query & Trend Engine]
I[Local Dashboard<br/>localhost only]
end
A --> D
B --> E
C --> D
D --> F
E --> F
F --> G
G --> H
H --> I
Roadmap
Contributing
Good first issues:
- Add Oura Ring CSV parser
- Add Fitbit CSV parser
- Add Garmin
.fit file support
- Add missing Whoop metrics to schema
- Improve test coverage
See good first issue labels to get started.
Built With AI Assistance
Leo Health was built with AI assistance and reviewed, tested, and maintained by @sandseb123. All security fixes, architectural decisions, and code reviews were done by a human. If you find issues or have questions, open a GitHub issue or reach out directly.
License
Leo Core — MIT. Free to use, modify, and distribute.
Leo Max (AI Coach) is a separate commercial product — join the waitlist.
Built by sandseb123
sandseb123/Leo-Health-Core