Overview
Runbooks
On-call playbooks for the incidents we've actually had.
Runbooks are short, prescriptive guides for specific incident classes. Each runbook assumes you’re on-call, just got paged, and need to stop the bleeding in minutes. No theory. No history. Read the symptom, follow the steps.
Pages in this section
- Langfuse alert response — what to do when the auto-eval loop fires a Slack alert.
More runbooks will land here as incident patterns emerge in production.
Was this helpful?