Thank you for the suggestion, I will explore this in the next iteration. I'm learning how to translate how humans do context management into how agents should do them