My Agent Forgot Everything Mid-Task: A Debugging Retrospective

An LLM agent lost critical decisions from its context window, causing inconsistent behavior. The fix was to log the full transcript, replay sessions, model trun