An AI-powered autonomous arbitrage system that finds and analyzes profitable opportunities across multiple platforms, generating revenue 24/7.

Arbi is a production-ready autonomous arbitrage system that uses AI to:

• 🔍 Find underpriced products across eBay, Amazon, Walmart, Target
• 🤖 Score opportunities with AI (0-100 point algorithm)
• 💰 Calculate real profit after fees, shipping, and costs
• ⚡ Execute trades automatically (with budget controls)
• 📊 Generate passive income for users
• 💵 Platform earns 25% commission on profits

Revenue Potential: $15k-40k/month with proper scaling

• Opportunity Analyzer - Scores each opportunity 0-100 points
• Risk Manager - Enforces budget limits and spending controls
• Confidence Scoring - Filters low-quality opportunities automatically
• Multi-Strategy System - eBay arbitrage, retail arbitrage, seasonal deals

1. eBay Scout - Finds items listed below sold price average (FREE)
2. Web Scraper - Scrapes Target, Walmart, eBay with Playwright (FREE)
3. Rainforest Scout - Gets Amazon data without Amazon API ($49/mo)

• Per-opportunity spending limits ($400 default)
• Daily spending limits ($1,000 default)
• Monthly budget caps ($10,000 default)
• Risk tolerance settings (conservative/moderate/aggressive)
• Real-time spending tracking

GET  /api/arbitrage/opportunities       - Find current opportunities
GET  /api/arbitrage/opportunities?minProfit=10&minROI=15
GET  /api/arbitrage/health              - System health check
POST /api/arbitrage/execute             - Execute an opportunity
GET  /api/arbitrage/settings            - Get user settings
PUT  /api/arbitrage/settings            - Update budget/risk settings

• Node.js v18+
• pnpm v8+
• eBay Developer API key (FREE - get at developer.ebay.com/join)

# Clone the repository
git clone https://github.com/ActivateLLC/arbi.git
cd arbi

# Install dependencies
pnpm install

# Build all packages
pnpm build

# Configure environment
cp .env.example .env
# Add your eBay API key: EBAY_APP_ID=your_app_id_here

# Start the API server
cd apps/api
node dist/index.js

# Check system health
curl http://localhost:3000/api/arbitrage/health

# Find opportunities
curl http://localhost:3000/api/arbitrage/opportunities

# Filter by profit/ROI
curl "http://localhost:3000/api/arbitrage/opportunities?minProfit=20&minROI=15"

{
  "totalFound": 15,
  "recommended": 8,
  "opportunities": [
    {
      "opportunity": {
        "title": "Apple AirPods Pro (2nd Gen)",
        "buyPrice": 189.99,
        "sellPrice": 249.99,
        "estimatedProfit": 19.50,
        "roi": 10.26,
        "buySource": "Target Clearance",
        "sellSource": "eBay"
      },
      "analysis": {
        "score": 72,
        "shouldExecute": true,
        "reasons": ["High confidence based on historical data"]
      },
      "riskAssessment": {
        "approved": true,
        "budgetCheck": { "passed": true }
      },
      "recommended": true
    }
  ]
}

arbi/
├── apps/
│   ├── api/              # Main API service (Express + TypeScript)
│   └── web/              # Customer-facing React app
├── packages/
│   ├── arbitrage-engine/ # 🎯 CORE: Autonomous arbitrage system
│   │   ├── scouts/       # Data source integrations
│   │   ├── analyzer/     # AI opportunity scoring
│   │   └── risk-manager/ # Budget & risk controls
│   ├── ai-engine/        # OpenAI Agents SDK integration
│   ├── web-automation/   # Playwright browser automation
│   ├── voice-interface/  # Whisper + ElevenLabs
│   ├── transaction/      # Hyperswitch payment processor
│   └── data/             # PostgreSQL + Redis
├── scripts/
│   └── get-ebay-api-key.ts # Automated API key creation
└── docs/
    ├── DEPLOY_NOW.md             # 3 deployment options (2-5 min)
    ├── LAUNCH_CHECKLIST.md       # Complete setup guide
    ├── AMAZON_API_ALTERNATIVES.md # API solutions
    ├── QUICKSTART_EBAY.md        # eBay setup (5 min)
    └── ENHANCEMENT_ROADMAP.md    # Future ML/RL improvements

• Set budget limits (daily, monthly, per-opportunity)
• System finds profitable opportunities automatically
• Execute trades manually or automatically
• Keep 75% of all profits

• Earn 25% commission on all profits
• No risk - users provide capital
• Scales infinitely with user base
• Recurring revenue from successful trades

Per user. Scale to 100 users = $112k-1.35M/month platform revenue

npm i -g @railway/cli
railway login
railway init
railway up
railway variables set EBAY_APP_ID=your_app_id_here

1. Connect your GitHub repo at render.com
2. Set build command: pnpm install && pnpm build
3. Set start command: cd apps/api && node dist/index.js
4. Add environment variable: EBAY_APP_ID

docker build -t arbi .
docker run -p 3000:3000 -e EBAY_APP_ID=your_key arbi

See DEPLOY_NOW.md for detailed deployment guides

# Required
EBAY_APP_ID=your_ebay_app_id        # Get at developer.ebay.com/join (FREE)

# Optional (for additional data sources)
RAINFOREST_API_KEY=your_key         # Amazon data API ($49/mo, 1000 free)
OPENAI_API_KEY=your_key             # For voice features

# Server
NODE_ENV=production
PORT=3000

Configure in code or via API:

{
  dailyLimit: 1000,           // Max spend per day
  perOpportunityMax: 400,     // Max per single trade
  monthlyLimit: 10000,        // Max spend per month
  reserveFund: 1000,          // Emergency reserve
  riskTolerance: 'moderate',  // conservative | moderate | aggressive
  enabledStrategies: [
    'ecommerce_arbitrage',
    'seasonal_arbitrage',
    'clearance_arbitrage'
  ]
}

• DEPLOY_NOW.md - Deploy in 2-5 minutes (Railway/Render/VPS)
• LAUNCH_CHECKLIST.md - Complete setup walkthrough
• AMAZON_API_ALTERNATIVES.md - 3 solutions without Amazon API
• QUICKSTART_EBAY.md - Get eBay API key in 5 minutes
• ENHANCEMENT_ROADMAP.md - ML/RL improvements (3-5x profit)

# Install dependencies
pnpm install

# Build all packages
pnpm build

# Run specific package
pnpm --filter @arbi/api dev
pnpm --filter @arbi/web dev

# Run tests
pnpm test

# Lint
pnpm lint

# Type check
pnpm type-check

• Scouts scan eBay, retail sites, APIs every 60 seconds
• Extract product prices, sold prices, demand metrics
• Cache results to reduce API calls

// AI Scoring Algorithm (0-100 points)
score =
  + profitPotential (0-30 points)  // ROI-based
  + confidence (0-25 points)        // Historical data
  + speedToProfit (0-20 points)     // Time to sell
  + riskLevel (0-15 points)         // Low/medium/high
  + volatility (0-10 points)        // Price stability

• Check budget limits (daily, monthly, per-opportunity)
• Verify spending capacity
• Calculate risk score
• Approve or reject opportunity

• Purchase item from buy source
• Create listing on sell platform
• Track inventory and sales
• Calculate actual profit
• Update ML models with results

• Budget limits enforced at multiple levels
• Spending tracked in real-time
• Risk tolerance configurable per user
• All API calls rate-limited
• Secure environment variable management
• No sensitive data in logs

• Deploy to Railway/Render
• Enable eBay scout
• Target 10-20 users
• Revenue: $5k-10k/month

• Add web scraper scout
• Implement auto-execution
• Scale to 100 users
• Revenue: $50k-100k/month

• Add ML price prediction
• Implement reinforcement learning
• Add more data sources
• Scale to 500 users
• Revenue: $250k-500k/month

1. Retail Arbitrage - Buy clearance, sell at market price
2. eBay Flipping - Find underpriced listings, resell
3. Seasonal Trading - Buy off-season, sell peak season
4. Brand Arbitrage - Regional price differences
5. Liquidation - Bulk purchases from liquidation auctions

• Core arbitrage engine
• eBay API integration
• Web scraping system
• AI opportunity scoring
• Risk management
• REST API
• Auto-execution system
• ML price prediction (TensorFlow.js)
• Reinforcement learning (PyTorch + RLlib)
• Mobile app
• Multi-user platform
• Social features (share opportunities)

See ENHANCEMENT_ROADMAP.md for detailed technical roadmap.

• Backend: Node.js, TypeScript, Express
• Frontend: React, Vite, TailwindCSS
• Data: PostgreSQL, Redis
• AI: OpenAI Agents SDK, Custom ML models
• Automation: Playwright
• APIs: eBay Finding API, Rainforest API
• Infrastructure: Docker, Railway, Render

Traditional arbitrage requires:

• ❌ Manual product research (hours per day)
• ❌ Constant price monitoring
• ❌ Spreadsheet profit calculations
• ❌ Risk of bad deals
• ❌ Limited scaling

Arbi automates everything:

• ✅ AI finds opportunities 24/7
• ✅ Instant price comparisons
• ✅ Automatic profit calculations
• ✅ Risk management built-in
• ✅ Infinite scaling potential

• Issues: GitHub Issues
• Docs: See /docs folder
• Email: [email protected]

MIT License - See LICENSE file

Built with ❤️ by ActivateLLC

Transform market inefficiencies into automated revenue streams.