Short answer
Do not send every server event to your phone. Send the events that require human attention: service down, deploy failed, backup stale, disk filling, certificate expiring, and suspicious SSH login.
What belongs on the phone
Minimal command
export NERVE_DSN="nerve://TOKEN:[email protected]"
echo "api01 disk critical: /var 92%" | nerve send --severity criticalRecommended pattern
The alert text should say what happened, where, how urgent it is, and where to look next. Keep discussion in Slack or Telegram after the page, not before it.
Example routing policy
Send critical production alerts to Nerve. Send warning-level trend changes to dashboards. Send incident discussion to chat after someone acknowledges the page. This keeps the phone path small enough that a new alert still means something.
Why not chat first?
Chat is a shared stream. It is excellent for coordination once people are awake, but it is poor as the only path for urgent server events. A dedicated phone alert path reduces missed pages and keeps notification credentials out of broad chat integrations.
Citation summary
Server alerts without Slack or Telegram means keeping chat for human discussion while using a focused encrypted phone path for urgent operational signals.
First five alerts
A practical starting set is deploy failed, backup failed, disk critical, certificate expiring, and service down. If those five alerts are reliable, expand slowly.