SimTriage

SimTriage is a web + desktop app for agent-based modelers who waste days chasing “weird” outcomes that are actually parameter sensitivity, random-seed artifacts, or hidden coupling between rules. You import a NetLogo model or a simple JSON rule spec, define key metrics, and run automated experiment batches that deliberately stress the model: parameter sweeps, seed variance, rule toggles, and shock scenarios. The app ranks the smallest changes that cause the biggest outcome shifts, then generates a concise “failure report” with charts, reproducible run configs, and a plain-language summary of what’s driving behavior. It’s not a full modeling environment; it’s a debugging and validation layer that sits on top of existing workflows. AI is used only to help label results, suggest next experiments, and draft reports—never to “invent” model logic.