What problem does it solve?
This Skill provides a systematic, authoritative playbook to diagnose and triage Azure production failures by collecting read-only evidence from resource health, logs, metrics, detectors, and Kubernetes state so engineers can identify probable root causes and next remediation steps quickly and safely.
Core Features & Use Cases
- Service-wide diagnostics: Identify failing Azure resources and surface resource health, activity logs, and service incidents.
- Application-level troubleshooting: Query Application Insights and Log Analytics using KQL to find errors, failed requests, and dependency failures.
- Container and Kubernetes support: Investigate Container Apps and AKS clusters, including image pull failures, CrashLoopBackOff, pending pods, CoreDNS/DNS issues, node NotReady, and ingress or load balancer failures.
- MCP & CLI guidance: Prefer AppLens / AKS-MCP detector flows and fall back to az and kubectl with clear, read-only fallbacks and explicit guardrails for any destructive actions.
- Use case: When an AKS cluster reports widespread pod failures, use this Skill to gather Azure-side evidence, run targeted kubectl reads, run KQL to correlate telemetry, produce a failure domain, confidence level, and recommended next steps.
Quick Start
Use the azure-diagnostics skill to perform a read-only investigation of AKS cluster 'my-cluster' in resource group 'my-rg' for CrashLoopBackOff pods and return evidence, likely root cause, confidence, and remediation options.
