An Amazon Deal Tracker With No Backend, Where Your Browser Scrapes
A real-time tracker for used and refurbished GPU, RAM, and SSD deals on Amazon, built with client-side scraping and a 97-line Cloudflare Worker.
sortsafe is an Amazon deal sniper aimed at PC parts. It tracks real-time pricing on GPUs, RAM, and SSDs, and it pays attention to the listings most trackers ignore: used, refurbished, and Amazon Warehouse. Price history is Keepa-driven, so a "deal" is judged against where the item has actually sat over time. You can use it at sortsafe.com.
No backend to run
The defining decision is that there is no backend doing the scraping. Your browser does it. That keeps the operating cost near zero and means every user is scraping on their own behalf rather than through a shared server that Amazon can throttle in one shot. The catalog lives locally in IndexedDB, so your tracked parts and history persist in the browser instead of a database I have to host and maintain.
Ninety-seven lines against the bot wall
Browsers cannot fetch Amazon HTML cross-origin directly, so there is one small piece of server code: a 97-line Cloudflare Worker that proxies the HTML through. The clever bit is how it handles Amazon's bot wall. Instead of parsing for a captcha page, the Worker detects the robot check by response size. If the response comes back under 50 KB, it is almost certainly a bot challenge rather than a real product page, so the Worker retries. It is a crude signal, but it is fast, cheap to evaluate, and it holds up in practice.
The whole thing is a study in doing more with less: push the heavy work to the client, keep the shared surface tiny, and lean on a blunt heuristic where a blunt heuristic is enough.
Have something that needs building, or stabilizing?
These notes are the work log. The paid work runs through Moonshine Labs, my product and engineering studio. Tell us what you're building, or grab a call.
Work with us →