| name | devopsguru-diagnostics |
| version | 1.0.0 |
| last_updated | 2025-04-12 |
| description | Use this skill to investigate and troubleshoot Amazon DevOps Guru problems by analyzing insights, anomaly detection, resource coverage, notification channels, cost estimation, proactive insights, and following structured runbooks. Activate when: missing insights, false anomalies, resource coverage gaps, notification failures, cost estimation issues, or the user says something is wrong with DevOps Guru.
|
| compatibility | Requires AWS CLI with devops-guru, CloudWatch, SNS, SSM, CloudFormation, and IAM permissions.
|
Amazon DevOps Guru Diagnostics
When to use
Any Amazon DevOps Guru investigation — insights, anomaly detection, resource coverage, notifications, cost estimation, or proactive insights.
Investigation workflow
Step 1 — Collect and triage
aws devops-guru describe-account-health
aws devops-guru describe-account-overview --from-time <epoch>
aws devops-guru list-insights --status-filter ONGOING
aws devops-guru describe-resource-collection-health --resource-collection-type AWS_SERVICE
Step 2 — Domain deep dive
aws devops-guru describe-insight --id <insight-id>
aws devops-guru list-anomalies-for-insight --insight-id <insight-id>
aws devops-guru list-recommendations --insight-id <insight-id>
aws devops-guru list-notification-channels
Step 3 — Detailed investigation
aws devops-guru describe-organization-health
aws devops-guru get-cost-estimation
aws devops-guru list-events --filters '{"ResourceCollection":{"CloudFormation":{"StackNames":["<stack>"]}}}'
Read references/guardrails.md before concluding on any DevOps Guru issue.
Gotchas: Amazon DevOps Guru
- DevOps Guru needs time to learn normal behavior. Anomalies appear after baseline period.
- Resource coverage must be explicitly configured (CloudFormation stacks, tags, or account-wide).
- Insights combine multiple anomalies. A single insight may have many related anomalies.
- Proactive insights predict issues before they occur. They require sufficient historical data.
- Cost estimation is approximate. Actual costs depend on resource count and activity.
- SNS notifications must be configured for alerts. No default notification channel.
Anti-hallucination rules
- Always cite insight IDs, anomaly details, or API responses as evidence.
- DevOps Guru needs learning time. Never expect instant anomaly detection.
- Resource coverage must be configured. Never assume all resources are monitored.
- Spend no more than 2 minutes on any single hypothesis. Pivot if inconclusive.
14 runbooks
| Category | IDs | Covers |
|---|
| A — Insights | A1-A3 | Reactive insights, proactive insights, insight quality |
| B — Anomalies | B1-B2 | Anomaly detection, false positives |
| C — Coverage | C1-C2 | Resource coverage, stack configuration |
| D — Notifications | D1-D2 | SNS channels, notification filtering |
| E — Cost | E1-E2 | Cost estimation, cost optimization |
| F — Proactive | F1-F2 | Proactive insights, recommendations |
| Z — Catch-All | Z1-Z2 | General troubleshooting, IAM issues |
