This looks interesting. If I’m understanding, it takes the raw html elements of a page and saves them into a RAG? Is this because feeding the whole page into an LLM prompt would be bigger than the content window?

Also love that it has both backend source code and a service available