Alex Getman
1.7 KiB
miband-bot
Self-hosted Telegram bot for Xiaomi Fitness and Mi Band health data.
It synchronizes steps, sleep, heart rate, SpO2, stress, daily activity, weight, and workouts into local SQLite files and lets you view or export them from Telegram. No third-party data service is involved.
Designed for one private owner. Not a public bot or a medical service.
Features
- Daily health summaries, history, weekly reports, trends, family stats, and comparisons.
- Xiaomi QR login and scheduled or manual synchronization.
- CSV/ZIP export from Telegram.
- English interface by default, with Russian and Spanish switchers in Settings.
- Bun/TypeScript runtime with Docker Compose.
Quick start
cp .env.example secrets.env
# Set TELEGRAM_BOT_TOKEN in secrets.env
bun install
bun run check
docker compose up -d --build
Runtime data is stored in ./data. Keep secrets.env, data/, Xiaomi tokens,
SQLite databases, and exports private.
Development
bun install
bun run check
bun run dev
The service exposes /healthz and /readyz on port 8080.
Production
The VM-106 deployment uses the existing /opt/miband-tracker directory and the
compose.vm106.yaml file. Production images are published to GHCR and activated by
.github/workflows/cd.yml as immutable digests. Enable its deploy job with repository variable
DEPLOY_ENABLED=true and the secrets DEPLOY_HOST, DEPLOY_USER, DEPLOY_SSH_PRIVATE_KEY, and
DEPLOY_KNOWN_HOSTS. Keep /opt/miband-tracker/secrets.env and /opt/miband-tracker/data on the
server; the workflow does not copy either one.