AI-powered crypto assistant that monitors the market, analyzes technical indicators, and provides intelligent, explainable trading insights. Designed for accessibility, it offers real-time updates, charts, and predictions to help users make informed decisions.
Note: Hosted on Render free tier. The app may take 1-2 minutes to wake up after inactivity. Data refreshes automatically on startup.
Authentication Required: The app requires authentication. You can create a free account to explore all features. You can delete your account at any time from the Account page.
graph TB
User[User Browser] --> Frontend[React Frontend<br/>Render Static Site]
Frontend --> Backend[Flask Backend<br/>Render Docker]
Backend --> DB[(PostgreSQL<br/>Render)]
Backend --> Binance[Binance API<br/>Real-time Prices + OHLCV]
Backend --> CoinGecko[CoinGecko API<br/>Metadata + Market Cap]
Backend --> FearGreed[Alternative.me API<br/>Fear & Greed Index]
Backend --> Groq[Groq LLM<br/>llama-3.3-70b-versatile]
Backend -.WebSocket.-> Frontend
- Python 3.12+, Flask, SQLAlchemy, Flask-Migrate
- Flask-SocketIO (WebSocket for real-time updates)
- gevent (async worker for Socket.IO)
- JWT (Flask-JWT-Extended), Bcrypt (password hashing)
- Resend (email verification & password reset)
- pandas-ta (technical indicators), Matplotlib (chart generation)
- NumPy, Pandas (data processing)
- React 18, Vite, Tailwind CSS
- TradingView Lightweight Charts (interactive price charts)
- Lucide React (icon library)
- Axios (HTTP client), Socket.IO Client (WebSocket)
- PostgreSQL (production on Render)
- SQLite (local development)
- Groq LLM (
llama-3.3-70b-versatilemodel for AI predictions)
- Binance API – Real-time prices and OHLCV historical data
- CoinGecko API – Coin metadata, market cap, global volume
- Alternative.me API – Fear & Greed Index
- Docker (backend containerization)
- GitHub Actions (CI/CD pipeline)
- Render (hosting: Docker backend, static frontend, PostgreSQL)
- AI Predictions with natural language reasoning (Buy/Sell/Hold recommendations)
- Real-time WebSocket Dashboard with live price updates every minute
- Technical Indicators: RSI, MACD, Bollinger Bands, Stochastic RSI, SMA, EMA
- Trading Charts with candlesticks, support/resistance levels, volume overlay
- Fear & Greed Index from Alternative.me (historical data + current value)
- Top 24h Volume with sparkline charts showing price trends
- Coin Detail Pages with TradingView charts and coin descriptions
- JWT Authentication with email verification via Resend
- Favorite Coins – Save and manage preferred cryptocurrencies per user
- Account Management – Edit username, delete account
- Password Reset via email with secure token-based flow
- Search Bar for quick coin lookup in navbar
- Coin Page Dropdown – Quick navigation to any coin from navbar
- Mobile-Responsive design with dark/light mode support
- Subtle Background Animation on auth pages (Ken Burns effect)
- Live Updates via WebSocket with REST API fallback
- Loading States and error handling throughout
-
Data Collection
- Fetches OHLCV (Open, High, Low, Close, Volume) data from Binance API
- Retrieves Fear & Greed Index from Alternative.me
- Queries historical data from PostgreSQL database
-
Technical Analysis
- Calculates indicators using
pandas-ta:- Trend: SMA (50, 200), EMA (50, 200)
- Momentum: RSI (14), MACD + Signal Line, Stochastic RSI (K%, D%)
- Volatility: Bollinger Bands (upper, middle, lower)
- Computes support and resistance levels from recent price windows
- Calculates indicators using
-
AI Prediction
- Sends all indicators + market data to Groq LLM (
llama-3.3-70b-versatile) - Receives Buy/Sell/Hold recommendation with natural language explanation
- Two report modes:
- Concise: Quick recommendation (500 tokens)
- Full: Detailed analysis with reasoning (2500 tokens)
- Sends all indicators + market data to Groq LLM (
-
Visualization
- Generates Matplotlib charts:
- Price chart with candlesticks + SMA/EMA overlays
- MACD/RSI chart with signal zones
- Bollinger Bands chart with price action
- Returns PNG images via Flask endpoints
- Generates Matplotlib charts:
- Python 3.12+ with pip
- Node.js 18+ with npm
- Git
git clone https://github.com/Ell-716/BASI-Crypto-Agent.git
cd BASI-Crypto-Agent# Create and activate virtual environment
python3 -m venv .venv
source .venv/bin/activate # On Windows: .venv\Scripts\activate
# Install dependencies
pip install -r requirements.txt
# Set up environment variables
cp .env.example .env
# Edit .env and fill in your API keys (see Environment Variables section)
# Run database migrations
flask db upgrade
# Backfill historical data (required on first run)
python backfill.py
# Start the backend server
python app.pyImportant Notes:
- Never use
flask run— always usepython app.py(Socket.IO requires gevent worker)- VPN must be off when running locally (Binance API may block VPN traffic)
- Backend runs on
http://localhost:5050by default
# In a separate terminal
cd frontend
npm install
npm run devFrontend runs on http://localhost:5173 by default.
Create a .env file in the project root with the following variables:
# Flask & JWT Configuration
SECRET_KEY=your-secret-key-here # Flask secret key (generate with: python -c "import secrets; print(secrets.token_hex(32))")
JWT_SECRET_KEY=your-jwt-secret-key-here # JWT signing key (generate with same command)
# API Keys
GROQ_API_KEY=your-groq-api-key # Get from: https://console.groq.com/
RESEND_API_KEY=your-resend-api-key # Get from: https://resend.com/
# Database
DATABASE_URL=sqlite:///crypto_agent_dev.db # Local: SQLite, Production: PostgreSQL URL from Render
# URLs
FRONTEND_URL=http://localhost:5173 # Frontend URL (production: https://app.basiai.org)
BACKEND_URL=http://localhost:5050 # Backend URL (production: https://api.basiai.org)
# External APIs
BINANCE_BASE_URL=https://api.binance.com # Production on Render: https://data-api.binance.vision- Groq API: Free tier available at console.groq.com
- Resend: Free tier (100 emails/day) at resend.com
- Binance & CoinGecko: No API key required for public endpoints
The project includes 40 pytest tests covering:
- Authentication flow (register, login, email verification, password reset)
- AI prediction endpoints
- Coin data retrieval
- Environment configuration
- Database models
# Activate virtual environment
source .venv/bin/activate
# Run all tests
pytest -v
# Run specific test file
pytest tests/test_auth.pyTests use an in-memory SQLite database with mocked external API calls (Binance, CoinGecko, Groq).
- Workflow:
.github/workflows/ci.yml - Triggers: Every push to
mainand all pull requests - Steps:
- Checkout code (
actions/checkout@v6) - Set up Python 3.12 (
actions/setup-python@v6) - Install dependencies
- Run full test suite with pytest
- Deploy to Render automatically on merge to
main
- Checkout code (
- Backend: Deployed as Docker container on Render (auto-deploy from
main) - Frontend: Deployed as static site on Render (auto-deploy from
main) - Database: Managed PostgreSQL on Render
POST /users/add_user– Register new userPOST /users/login– Login and receive JWT tokensGET /users/verify?token=<token>– Verify email addressPOST /users/resend-verification– Resend verification emailPOST /users/refresh– Refresh access tokenPOST /users/request-password-reset– Request password reset linkPOST /users/reset-password– Reset password with token
GET /api/coins– List all supported coins (ordered by ID)GET /api/coins/<coin_id>– Get coin by IDGET /api/coins/<coin_id>/history– Get historical price data
GET /dashboard/fear-greed– Current Fear & Greed IndexGET /dashboard/top-volume– Top coins by 24h volume with sparklinesGET /dashboard/sparkline/<symbol>– Sparkline chart data for specific coinGET /dashboard/snapshot/<symbol>– Market snapshot for specific coinGET /dashboard/coins– Top 10 coins data
GET /predict?coin=<symbol>&timeframe=<1h|1d|1w>&report_type=<concise|full>– Get AI prediction- Returns: JSON with analysis text, charts (base64 encoded)
GET /chart/price/<symbol>?timeframe=<1h|1d|1w>– Price chart PNGGET /chart/macd-rsi/<symbol>?timeframe=<1h|1d|1w>– MACD/RSI chart PNGGET /chart/bollinger/<symbol>?timeframe=<1h|1d|1w>– Bollinger Bands chart PNG
GET /users/<user_id>– Get user profile (JWT required)PUT /users/<user_id>– Update username and favorite coins (JWT required)DELETE /users/<user_id>– Delete account (JWT required)
connect– Client connects, receives initial coin datacoin_data– Server emits live price updates every minuterequest_coin_data– Client can request immediate data updatedisconnect– Client disconnects
-
Backend: Docker deployment
- Build command:
docker build -t basi-backend . - Start command:
python app.py - Port: 5050
- Environment:
BINANCE_BASE_URL=https://data-api.binance.vision
- Build command:
-
Frontend: Static site
- Build command:
cd frontend && npm install && npm run build - Publish directory:
frontend/dist
- Build command:
-
Database: PostgreSQL (Render managed service)
- App spins down after 15 minutes of inactivity
- Cold start: ~1-2 minute delay on first request after spin-down
- Data refresh: Automatic on cold start if data is older than 6 hours
- Historical data: 6-hour staleness threshold
- CoinGecko snapshots: 24-hour staleness threshold
- Fear & Greed Index: 24-hour staleness threshold
The codebase includes cron scripts for production use:
# backend/cron_update.py - Update historical data & indicators
# Run hourly: 0 * * * *
# cron_update_fgi.py - Update Fear & Greed index
# Run daily at 04:05: 5 4 * * *
# cron_update_top_volume.py - Update top volume
# Run daily at 02:15: 15 2 * * * *
# cron_update_snapshot.py - Update Market cap and Volume
# Run daily at 02:00: 0 2 * * *Note: Cron jobs are not needed on Render free tier due to automatic data refresh on cold start. They are designed for paid hosting with always-on backends.
- Cold start delay: 1-2 minutes after 15 min inactivity
- WebSocket intermittency: May disconnect during periods of low activity
- Rate limiting: Shared IP may be rate-limited by external APIs (Binance, CoinGecko)
- 1-hour predictions require historical data less than 6 hours old
- Weekly predictions always fetch fresh data from Binance (no DB dependency)
- Verification and password reset emails sent from custom domain:
noreply@basiai.org
- VPN users: Binance API blocks some VPN IPs (use
BINANCE_BASE_URL=https://data-api.binance.visionworkaround) - Geo-restrictions: Some regions may be blocked by Binance
- Migrate to FastAPI – Improve async performance and API documentation
- Portfolio Tracking – Track holdings, profit/loss, realized gains
- Price Alerts – Email/push notifications for price targets
- Strategy Layer – Combine predictions with risk management (stop-loss, take-profit)
- Historical Backtesting – Test AI predictions against past data
- Multi-LLM Support – Compare predictions from different models (GPT-4, Claude, etc.)
- Mobile App – React Native version with push notifications
This project is licensed under the MIT License — see the LICENSE file for details.
Disclaimer: This project is for educational and demonstration purposes only. It is not financial advice. Cryptocurrency trading involves substantial risk of loss. Always do your own research before making investment decisions.
Created with passion by Elena Bai
- GitHub: @Ell-716
- Email: elenabai.2021@gmail.com
Copyright © 2025-2026 Elena Bai. All rights reserved.
⭐ If you find this project helpful, please give it a star on GitHub! ⭐