Interesting... how did you do the scraping of the documentation?
I haven't started digging into the RAG performance, so open to ideas on ways to improve this!
I’ll add an issue for this though since I think it could definitely improve the performance!