I built my own tool for that: files-to-prompt to chuck all of the files in a directory into a single stream, then I pipe that to my LLM tool and send it to a large context model like Claude 3 Opus to ask questions about it

→ View original post on X — @simonw