Pod Management Skill

This skill provides direct access to the Kubernetes cluster's pod lifecycle and operational data. It abstracts the kubectl complexity into high-signal tools for agentic analysis.

🛠 Available Tools

Tool	Purpose	Input Schema
inspect_pods	List all pods in a namespace with status summary	{ "namespace": "string" }
describe_pod	Get detailed status and configuration for a pod	{ "namespace": "string", "name": "string" }
list_pod_events	List all lifecycle events associated with a pod	{ "namespace": "string", "name": "string" }
get_pod_logs	Retrieve logs from a specific pod/container	{ "namespace": "string", "name": "string", "container": "string", "tail_lines": number, "previous": boolean }
delete_pod	Delete a specific pod (useful for restarts)	{ "namespace": "string", "name": "string", "grace_seconds": number }

📋 Standard Workflows

1. Pod Health Check

When a service is reported as "Down" or "Degraded":

1. Run inspect_pods in the relevant namespace to check for Pending, Failed, or CrashLoopBackOff statuses.
2. Use describe_pod to identify resource constraints (CPU/Memory limits) or configuration errors.
3. Check list_pod_events for recent BackOff or FailedScheduling events.

2. Log Analysis

If a pod is running but behaving incorrectly:

1. Run get_pod_logs with a reasonable tail_lines (e.g., 100).
2. If the pod has restarted, use previous: true to see the logs from the crashed container.

💡 Operational Tips

• Namespace: Most hub services live in the default or observability namespaces.
• Graceful Deletion: Use delete_pod only when a restart is necessary to clear a stuck state.

---

For detailed API documentation, see references/api-specs.md.