Observability incident triage
Skill jukrap/ai-agent-playbook/skills/devops/observability-incident-triage
Reusable AI agent skills, project templates, and guardrails for safer software maintenance and delivery.
npx -y skills add jukrap/ai-agent-playbook --skill observability-incident-triageAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
One thing to look at
- 2 stars2 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.
What its author says it does
Copied from the file, not written here
Use when triaging incidents, production errors, alerts, latency, error rates, queue backlogs, job failures, logs, metrics, traces, or post-incident runbooks.
SKILL.md
1.0 KB, 162 tokens by cl100k_base, as published. Nobody here has run it
Observability Incident Triage
Use this as the primary DevOps skill for active incidents and production signal review.
Workflow
- Capture the symptom, start time, impacted users/services, severity, recent changes, and available logs/metrics/traces.
- Prefer containment and rollback when user impact is active; do not delay mitigation for perfect root cause.
- Correlate signals across deploys, dependencies, queues, jobs, resources, and error boundaries before changing code.
- Verify recovery with the same signal that showed the incident, then record follow-up actions and durable runbook updates.
Reference
Read references/incident-evidence.md for active incident intake, containment, and handoff.
Read references/logs-metrics-traces.md when interpreting observability evidence.