PennyPincher

I built a Puppeteer-based scraper that adapts to messy site structures and dodges common anti-bot measures. Scraping is parallelized, with three workers running by default; a sweet spot I found between performance and system load. Users can adjust the number of workers if they want faster scans or lighter resource usage.

The app is split into layers: the Vue+Electron UI talks to a Node.js + Express backend through REST APIs. That backend is managed by a small C# WPF helper app I built, which makes sure Node and all required modules are installed, and starts the server silently on Windows startup so scraping always runs in the background. I chose this setup for flexibility: the UI can evolve independently, and the backend can eventually even be hosted remotely if needed.

On the front end, Electron provides a smooth cross-platform desktop experience. It combines the Vue-powered interface with Node.js under the hood, while also spinning up a small local server so you can access the UI from your phone on the same network; handy if you want to check prices from the couch.

For storage, I chose SQLite: lightweight, fast, and perfect for a local-first app. It keeps user data private on their machine while making lookups and alerts instant. I also used Electron's storage for caching favicons and product images, so the app looks polished without wasting bandwidth on repeated downloads.

Price alerts show up as native desktop notifications, and I added Telegram support for anyone who wants updates on the go. It's surprisingly satisfying to get a ping that a tracked product finally dropped in price.