AI Runbooks let you codify your incident response procedures so Nova can execute them automatically — or present them to on-call engineers with pre-filled commands and context. This turns tribal knowledge into repeatable, auditable workflows.
Go to Automation > Runbooks and click Create Runbook.
Give it a clear name like "Restart API Service on High Memory" and add a description of when this runbook should be used.
Each step can be one of three types:
Example runbook steps:
Link the runbook to specific alert types so it runs automatically. For example, trigger on alerts where service:api AND metric:memory.used_pct > 90.
Select Automatic to let Nova run the runbook without human intervention, or Guided to present steps to the on-call engineer for manual execution.
Start with Guided mode until you are confident the runbook works correctly. Once validated, switch to Automatic for faster remediation.
Use the Dry Run feature to simulate execution without affecting your infrastructure. Review the output of each step to make sure it behaves as expected.
Click Publish. Nova will track every execution with full logs, duration, and success/failure status in the Runbook History tab.