hassan-khan-web/Flux ? reverse-engineered prompt

Build me a web app and API that can take a search topic or URL and gather fresh information from the web, clean it up, remove noise and duplicates, and return a well structured result that AI tools can use right away.

I want it to be able to look up sources, scrape pages when needed, extract useful details like titles, authors, dates, and full text, then organize everything into clean markdown and JSON. If possible, add a simple scoring step so the best results are ranked higher, and cache repeated requests so it feels fast.

Please include a lightweight frontend where I can paste a query, see the collected sources, and view the cleaned output nicely. Also set it up so it can run locally with Docker, has clear API endpoints, and includes basic docs and tests. Use current docs online if you need to.