1. Receive the alert
Alertmanager, Zabbix, Sentry, LibreNMS or a generic webhook sends an alert to an IncidentRelay integration endpoint. Route intake tokens keep incoming traffic tied to explicit route configuration.
2. Match route and service
IncidentRelay evaluates source, team, labels and fields. The matched route determines ownership, grouping, default service, schedule or escalation policy, and outgoing channels.
3. Calculate who is on call
Rotations can include layers, membership periods and temporary overrides. IncidentRelay calculates the current responder and keeps calendar visibility available through internal views and calendar feeds.
4. Notify with context
Notifications include alert title, severity, status, source links and actions. Channels can include Mattermost, Telegram, email, browser push, webhooks and voice providers.
5. Act on the incident
Responders can acknowledge, resolve, silence, comment, request additional responders and keep stakeholders informed. Escalation policies continue only when the incident still needs attention.
6. Learn from the history
Alert groups retain status, labels, notifications and comments so the team can understand what happened after the alert is resolved.