An interactive incident-response simulator. Trigger a realistic outage across a simulated fifteen-service infrastructure, then investigate the logs, metrics and dependency map to find the root cause and restore service.
Everything on screen comes from one underlying model, so the story always holds together — the charts, the logs, the dependency map and the customer complaints agree because they're all computed from the same source rather than written separately. Failures spread the way they do in real systems: losing a cache makes things slow, losing a database makes them stop. One rule produces both, with no script per scenario.
Nothing is random, so an incident unfolds identically every time — which is also what lets an investigation survive a page refresh while storing almost nothing. Eight scenarios each teach a different shape of problem, and a wrong answer explains which evidence rules it out instead of just marking you incorrect. An optional guided mode walks newcomers through it, and any finished incident can be replayed second by second.