miband-bot

Русский | English

A personal self-hosted Telegram bot for your Xiaomi Fitness / Mi Band data.

Fetches steps, sleep, heart rate, SpO2, stress, daily activity, weight, and workouts from the Xiaomi Fitness cloud, stores them in a local SQLite database, and provides access to them directly from Telegram — without third-party services and without sharing your data with anyone.

This project is designed for a single owner. This is not a public bot or a medical service.

Features

• View recent steps, sleep, heart rate, SpO2, stress, weight, and workouts in Telegram.
• Manual and scheduled automatic synchronization.
• Auto-refresh of the pinned main menu message after background synchronization.
• History storage in SQLite (data/).
• Export of all tables to a ZIP archive with CSV files directly into the chat.
• Deployment via Docker Compose.
• Atomic writing of the Xiaomi token with 0600 permissions.
• Smart auto-binding to the first user (whitelist).

How it works

Mi Band → Xiaomi Fitness cloud → miband-bot → SQLite → Telegram / CSV

Docker Compose runs two processes:

• tracker — periodically synchronizes data from Xiaomi Fitness;
• fitness-bot — serves the Telegram menu, handles manual sync, and performs exports.

Both processes work with the same ./data folder. Concurrent write access is prevented by a file-based lock.

Requirements

• Docker and Docker Compose (or installed Python 3.11+).
• Telegram bot token from @BotFather.
• A Xiaomi account with Xiaomi Fitness data.

Quick Start

Method 1: Seamless One-Click Installation (Recommended)

If you don't have the project files on your machine yet, you can automatically download and set everything up using a single command in your terminal:

• macOS / Linux:

  curl -fsSL https://raw.githubusercontent.com/iAlexeyRu/miband-bot/main/install.sh | bash
  

• Windows (PowerShell):

  powershell -c "irm https://raw.githubusercontent.com/iAlexeyRu/miband-bot/main/install.ps1 | iex"
  

The installer will automatically create a miband-bot directory, download and extract the project files, verify dependencies, and launch the interactive setup! Running the same PowerShell command again on an already configured install will update the files and start the bot without asking for the Telegram token again.

---

Method 2: Launch from Downloaded Directory

If you have already cloned the repository via git clone or downloaded the ZIP archive manually:

• macOS / Linux:

  ./setup.sh
  

• Windows: Double-click the setup.bat file or run it in the console:

  setup.bat
  

The script will automatically check your environment, guide you step-by-step to get your Telegram bot token, create the secrets.env configuration, set up the Python virtual environment (if you choose to run without Docker), and let you launch the bot with a single key press! After setup, you can start the bot again with run_local.sh on macOS/Linux or run_local.bat on Windows from the miband-bot folder.

---

Method 3: Fully Manual Setup:

1. Copy the configuration template:

   cp secrets.env.example secrets.env
   

2. Specify your TELEGRAM_BOT_TOKEN in the secrets.env file. Leave the TELEGRAM_ALLOWED_USER_ID variable blank — the bot will automatically bind to you upon the first start.
3. Start the Docker containers:

   docker compose up -d --build
   

4. Open your created bot in Telegram and send the /start command — the bot will recognize your account, bind it as the sole owner, and begin synchronization!

Settings

All variables are in secrets.env:

Variable	Default	Description
TELEGRAM_BOT_TOKEN	—	Telegram bot token
TELEGRAM_ALLOWED_USER_ID	—	Allowed user ID (leave empty for auto-binding)
SYNC_INTERVAL	900	Background sync interval, in seconds
QUERY_DURATION	2	Fetch depth during sync, in days
ENABLE_FDS_SLEEP_DETAILS	true	Download detailed FDS night sleep data

Paths to the database and status files are defined in compose.yaml. For running without Docker, refer to secrets.env.example.

Data Files

Runtime files are created in ./data:

File	Content
token_<id>.json	Xiaomi auth token (secret)
miband_<id>.db	SQLite database with health data
status_<id>.json	Last sync status
allowed_user.id	ID of the bound owner
fitness_bot_state.db	Telegram menu internal state
sync_<id>.lock	Sync lock file

secrets.env, data/, *.db, token*.json, and status*.json are added to .gitignore — do not commit them.

Commands

Command	Action
/start	Open menu or start Xiaomi login flow
/sync	Start manual synchronization
/status	Show local database status

Local Development

python3 -m venv .venv
.venv/bin/pip install -r requirements-dev.txt -e mi-fitness-python
.venv/bin/python -m py_compile fitness_bot.py miband_sync.py \
    $(find miband_tracker -name '*.py' | sort)
.venv/bin/python -m pytest
.venv/bin/python -m pytest mi-fitness-python/tests/unit
.venv/bin/ruff check .
.venv/bin/python -m pip check

Entry points:

python -u miband_sync.py      # or: miband-sync
python -u fitness_bot.py      # or: miband-fitness-bot

Troubleshooting

The bot does not respond — check TELEGRAM_BOT_TOKEN, check the logs, and make sure you were the first to send /start to the bot to bind it. If you need to reset the bound owner, simply delete the file data/allowed_user.id and send /start again.

docker compose logs -f fitness-bot

Token not found — send /start and complete the Xiaomi login flow.

Token expired — start a re-login from the menu; the old file can be deleted from data/.

No SpO2 or sleep details — make sure this data is visible in the Xiaomi Fitness app itself. Availability depends on the band model and data sharing settings.

Everything broke after a Xiaomi update — this is an expected risk when working with unofficial APIs. Check issues and logs, then update the code or temporarily disable the problematic module.

Important: Reverse Engineering and Limitations

miband-bot is an unofficial project, not affiliated with Xiaomi, Zepp, Huami, or Telegram.

Data access is implemented via reverse engineering of closed APIs, therefore:

• Xiaomi may change the API without warning;
• authorization or synchronization may temporarily stop working;
• use this project only with your own accounts and data;
• comply with applicable laws and services' terms of use;
• wristband data is not a medical opinion.

License

The project is distributed under the GNU GPL v3.0 or later.

SDK mi-fitness-python is included as a vendored source copy under GNU GPL v3.0. Details are in VENDORED.md.