LLM Agent Planning under Deception

A collaborative effort to evaluate LLM agent architectures in deceptive text environments to understand the limits of multi-step planning and reasoning.