kagenti Kubernetes Debugging Skills
Skills for debugging and troubleshooting Kubernetes deployments.
Context-Safe Execution (MANDATORY)
All kubectl/oc commands MUST redirect output to files. Commands in this skill are shown in bare form for readability, but when executing them, always use this pattern:
# Set log directory (use cluster name or worktree to avoid session collisions)
export LOG_DIR=/tmp/rossoctl/k8s/${CLUSTER:-local}
mkdir -p $LOG_DIR
# Pattern: redirect output, return status
kubectl <command> > $LOG_DIR/<descriptive-name>.log 2>&1 && echo "OK" || echo "FAIL (see $LOG_DIR/<descriptive-name>.log)"
# When investigating failures: use Task(subagent_type='Explore') to read log files
# NEVER read large kubectl output directly into main conversation context
Available Sub-Skills
| Skill | Description |
|---|---|
k8s:pods |
Troubleshoot pod issues (CrashLoopBackOff, ImagePull, etc.) |
k8s:logs |
Query and analyze pod/container logs |
k8s:health |
Check platform health and component status |
k8s:live-debugging |
Iterative debugging on running clusters |
Quick Debugging
Check Pod Status
# All pods
kubectl get pods -A
# Failed pods
kubectl get pods -A --field-selector=status.phase!=Running,status.phase!=Succeeded
# Specific namespace
kubectl get pods -n team1
View Logs
# Agent logs
kubectl logs -n team1 deployment/weather-service --tail=100 -f
# Operator logs
kubectl logs -n rossoctl-system -l app=rossoctl-operator --tail=100
Check Events
kubectl get events -A --sort-by='.lastTimestamp' | tail -30
Platform Health
# All deployments
kubectl get deployments -A
# Services
kubectl get svc -A
# HTTPRoutes
kubectl get httproutes -A
Common Issues
- CrashLoopBackOff: Check logs, resource limits, configuration
- ImagePullBackOff: Check registry auth, image name
- Pending: Check resource requests, node capacity
- Evicted: Check disk pressure, memory limits