PageAlert

AI-powered website monitoring. Describe what you want, get notified when it appears.

pagealert.io

---

What is PageAlert?

PageAlert monitors any website using AI. Paste a URL, describe what you're looking for in plain English, and get notified via email when your conditions are met.

• Track product availability and restocks
• Monitor price drops across any store
• Watch for new listings on job boards, classifieds, or auction sites
• No CSS selectors or code required — just describe what you want

How it works

1. Paste a URL — any website with products, listings, or data you care about
2. Describe what you want — "MacBook Pro M4 Max under $3000 in black"
3. AI extracts the data — understands the page structure, finds items, generates match filters
4. Get notified — email alerts when your conditions are met, with direct links to matched items

Tech Stack

• Frontend: Next.js (App Router) + Tailwind CSS + shadcn/ui
• Backend: Convex (real-time database + scheduled functions)
• Auth: Better Auth with Google/GitHub OAuth
• Scraper: Playwright + Hono API (Dockerised)
• AI: Anthropic Claude for intelligent data extraction
• Billing: Polar for subscriptions
• Email: Resend for notifications
• Analytics: PostHog for product analytics + LLM observability
• Hosting: Vercel (web) + Railway (scraper)

Project Structure

prowl/
├── apps/
│   ├── web/              # Next.js app (frontend + Convex backend)
│   │   ├── app/          # Pages and routes
│   │   ├── components/   # UI components
│   │   ├── convex/       # Convex schema, mutations, queries, crons
│   │   ├── hooks/        # React hooks
│   │   └── lib/          # Utilities
│   └── scraper/          # Hono API + Playwright scraper (Docker)
│       └── src/
│           ├── routes/   # API endpoints
│           ├── services/ # Scraper + AI extractor
│           └── utils/    # URL validation, etc.
├── packages/
│   └── shared/           # Shared types and utilities
├── Dockerfile.scraper    # Docker build for scraper
├── docker-compose.yml    # Local development
└── turbo.json            # Turborepo config

Development

# Install dependencies
pnpm install

# Start the web app (port 3000)
cd apps/web && pnpm dev

# Start the scraper (port 3001) — needs Playwright browsers
cd apps/scraper && pnpm dev

# Or run both via Turborepo
pnpm dev

Environment Variables

Copy .env.example and fill in your values. See the infrastructure doc for details.

Features

• AI-powered extraction — Claude understands page structure and extracts structured data
• Smart matching — keyword filters, price ranges, blacklists
• Scheduled monitoring — automatic checks from every 5 minutes to every 24 hours
• Email notifications — match alerts and error notifications via Resend
• Change detection — tracks new items, removed items, and price changes between scans
• AI insights — confidence scores, match signals, and page accessibility notices
• Debug logs — full scrape history with raw AI responses for troubleshooting
• Multi-currency — detects and displays prices in the correct currency
• Subscription billing — Free, Pro, and Max tiers via Polar

License

MIT — see LICENSE for details.

Author

Built by James DiMonaco