[FEATURE]: __more context__ extract browser history & content summary

[ZaynJarvis]

📝 Feature Description [Please make everyone to understand it]

crop screen is one very important source, while since

1. resolution is compacted OCR is not exact, some context will lose.
2. some key info such as url, won't necessary be shown on screen.

I have this project before, it mimic Arc browser new tab feature. It use chrome extension to extract page url and page info, index in vectorDB, and features dense&sparse search for page retrieval. Could be a reference for browser context extraction.

https://github.com/ZaynJarvis/newtab

👍 What problem does this feature solve

new context

👾 What does the proposed API look like

--

🚑 Any additional information

No response

[ZaynJarvis]

newtab demo