Aviator Bot

An automated betting bot for the SportPesa Aviator game (powered by Spribe). It uses Playwright to drive a real Chrome browser, reads the crash history from the game iframe, and places bets on two independent panels based on configurable trigger conditions and recovery strategies.

The bot watches every Aviator round without betting. When a configured trigger condition fires — a high crash multiplier or a long streak of low crashes — it enters bet mode and places bets on both panels for up to N rounds. Each panel has its own cashout target and independent recovery logic. After a win or when all rounds are exhausted, it returns to watch mode.

WATCH -> TRIGGER -> BET (up to MAX_BET_ROUNDS) -> WIN or EXHAUST -> WATCH

P1 bet = (deficit + RECOVERY_PROFIT_TARGET) / PANEL1_CASHOUT

Panel 1 scales up each round to recover the cumulative deficit from previous losses. Panel 2 can be set to flat or independent recovery. With P1 assist enabled, P1 can also target a configured percentage of P2's deficit when P1 has no deficit of its own.

• Stop on Profit / Stop on Loss — global session limits in KES
• Burst cooldown — skip N watch rounds after each burst
• Stop on consecutive losses — halt session after N losing rounds in a row

Runs a single bot session in your terminal. Credentials and settings come from config.py.

source .venv/bin/activate
# Edit config.py with your credentials and preferred settings
python bot.py

A FastAPI server serves a web dashboard. Multiple users can log in with their own SportPesa credentials and run isolated bot sessions simultaneously.

source .venv/bin/activate
uvicorn server:app --host 0.0.0.0 --port 8000

Open http://localhost:8000 in a browser. Each user:

1. Enters their SportPesa phone number and password
2. Selects or creates a betting strategy
3. Clicks START BOT — a headless Chrome session starts in the background
4. Monitors live P&L, round count, and last event from the dashboard

Strategies are stored in strategies.json and managed from the dashboard.

• Global strategies (created by admin) — visible to all users, read-only for non-admins
• User strategies — created from the dashboard, visible and editable only by that user

Five strategies ship on first run:

AVIATOR/
├── bot.py                  <- Local CLI runner (reads config.py)
├── config.py               <- Settings for local bot
├── server.py               <- FastAPI server (web UI + REST API)
├── src/
│   └── bot.py              <- Server bot (per-session credentials, stop event)
├── static/
│   ├── index.html          <- Web dashboard
│   ├── help.html           <- Strategy guide
│   ├── how-it-works.html   <- Detailed mechanics documentation
│   ├── admin.html          <- Admin panel (strategy + user management)
│   └── img/                <- Screenshot assets for help page
├── requirements.txt        <- Local bot dependencies
├── requirements_server.txt <- Server dependencies
├── logs/                   <- Session log files (gitignored)
├── history/                <- Per-session CSV round history (gitignored)
└── strategies.json         <- Persisted strategy definitions (gitignored)

Two bot files: bot.py (local) and src/bot.py (server) are kept in sync. When changing strategy logic, update both.

• Python 3.11+
• Playwright (Chromium)
• FastAPI + Uvicorn (server mode)
• A SportPesa account (Kenya)

pip install -r requirements_server.txt
playwright install chromium

• Credentials are sent as plain JSON — run behind HTTPS (nginx + certbot) in production.
• The server holds sessions in memory; a restart clears all active sessions.
• RuntimeError: Event loop is closed in logs is harmless Playwright GC noise on Python 3.12.
• Demo mode reads balance from the Spribe game iframe instead of the SportPesa header. git add Dockerfile .dockerignore docker-compose.yml nginx/ .github/ git commit -m "Add Docker and CI/CD pipeline" git push origin production