Hazard detection is the act of determining that a hazardous condition exists now, usually by comparing a sensor reading against a defined limit.
Detection is the part of a safety system that must never fail, which is why it should depend on as little as possible - no network, no server, no external service. Everything else in a monitoring platform can degrade gracefully; detection cannot.
A sensor produces a value, the value is compared against a threshold, and a breach raises an alarm. The simplicity is the feature: the logic is auditable, reproducible and explainable after an incident.
A carbon monoxide reading crossing its limit triggers an audible alarm and a visual indication at the node.
Detection runs on the node itself, before any reading is published. The buzzer and display are driven by that local evaluation, so an outage costs visibility rather than the alarm.
