Thanks for the star and the question. Most tools that do search and summarisation either need custom selectors per site or pull in a full 200MB headless browser just to handle JS rendering. Focal Harvest does not. It runs on a cheap VPS, has no database, and works without any API key (If you want, you can provide an API key though).
The whole pipeline is plain Python with no browser dependency. JS-heavy pages are handled by extracting `__NEXT_DATA__`/`__NUXT_DATA__` JSON directly from the raw html, and wayback machine covers the rest. Everything saves as markdown and JSON files, which you can open by double-clicking. I've updated the repo now so that you can see the logs too.
The whole pipeline is plain Python with no browser dependency. JS-heavy pages are handled by extracting `__NEXT_DATA__`/`__NUXT_DATA__` JSON directly from the raw html, and wayback machine covers the rest. Everything saves as markdown and JSON files, which you can open by double-clicking. I've updated the repo now so that you can see the logs too.