alerting-engineering
Installation
SKILL.md
Alerting Engineering
Use this skill when
- Turning SLOs and metrics into alerts
- Setting severity, escalation, and routing
- Reducing alert noise
Do not use this skill when
- Building dashboards
- Debugging a system (incident playbook)
Instructions
- Alert on user impact symptoms and SLO burn, not every dip.
- Set paging thresholds so real incidents page; everything else goes to an issue.
- Include runbook links, severity, and owner in every alert.
- Use alerting on rate/trend, not raw thresholds alone.
- Keep a max alerts-per-person budget; retire noise.
- Escalate on silence or repetition.
- Post-incident, tune alerts as follow-up.