skip to content

Periscope gathers product and pricing information from interactive websites. The difficult part is not just reading a page: a useful result also needs to show where each fact came from so that someone can check it.

The system uses cloud browser sessions and Playwright automation to navigate sites, collect data, and attach source evidence. A web console provides a place to review the extracted results. The stack includes TypeScript, Next.js, and SQLite.

This is an engineering experiment in reliable extraction rather than a claim that every site can be parsed perfectly. The repository has the current implementation and setup details.