lucatrevisanii/multi-robot-coordination ? reverse-engineered prompt

Build me a browser based simulation of a swarm of robots that clear a field of tasks without any central planner.

Each robot should move around on a canvas, decide which task to go after by itself, and only share its bid and current winner info with nearby robots inside a communication range. Use that local consensus style auction so robots can still agree on task ownership even when the network is fragmented. I want the scene to update live, show the robots, tasks, and some simple stats like how many conflicts happened and how many tasks are left. It would be great if I can tweak the number of robots, tasks, and communication range and see how that changes coordination.

Please keep the logic clean enough that the simulation can also run without the browser for a quick invariant check. If you need to look up current docs online while building it, go ahead.