Outage Cascade Alert Routing
In telecom networks, a hardware failure (e.g., a cut fiber optic cable or switch failure) triggers a "cascade" of thousands of downstream alarm alerts (syslogs, pings failing) within seconds. If an autonomous troubleshooting agent is flooded with these alarms, it can become overwhelmed or enter infinite tool-calling loops.
How do you architect a multi-agent system to filter the noise and route the diagnostic workflow to the root-cause device?