1
Ask HN: Does piping LLM output into a RAG stack sound like a good idea?
I would like to know if this can improve agent memory or adherence to coding guidelines and policies in large codebases over time.
I'm also interested in this as a general concept.