Planning Against Nature develops advanced methods for planning and acting in dynamic environments where autonomous agents must remain safe under uncertain external events, including random acts of nature. A representative scenario is an Autonomous Underwater Vehicle (AUV) sampling objects of interest in busy shipping lanes, where every decision must balance mission progress with strict collision-avoidance guarantees throughout execution. The core objective is to generate sequential plans that are guaranteed to achieve the goal while ensuring the AUV will not collide with ships during execution, providing an important safety layer for longer-term autonomy with minimal human intervention, such as ocean or space exploration. To support both research and practical deployment, we are building PANSim, a visualization and simulation tool that makes safe planning strategies easier to explore, compare, and communicate across autonomous robotics, human-robot interaction, and cybersecurity use cases.