K8s Inspect Skill (Shell)

Execute a comprehensive Kubernetes cluster health inspection using shell scripting and kubectl, generating a dark-themed HTML dashboard report.

Overview

This skill enables agents to autonomously:
- Check Node Health — Readiness, CPU/Memory usage (via metrics-server), kubelet version, kernel version, uptime
- Analyze Pod Status — Running/Pending/CrashLoopBackOff/OOMKilled counts, abnormal pod listing with images and restart counts
- Audit Resource Quotas — ResourceQuota usage, PVC status and capacity, StorageClass binding
- Verify Certificates — kubeadm certificates expiration, Ingress TLS secrets expiration with countdown
- Inspect Network — CoreDNS, kube-proxy, Calico/Flannel/Cilium replica health
- Extract Warning Events — Recent Warning events across all namespaces
- Generate Reports — Compile all data into a polished dark-themed HTML dashboard

Prerequisites

1. kubectl installed and on PATH (v1.20+)
2. bash, bc, awk (pre-installed on most systems)
3. kubeconfig file with cluster credentials
4. metrics-server (optional, for CPU/Memory usage stats)

Quick Start

# Basic usage (uses default ~/.kube/config)
bash k8s_inspect.sh

# Specify a custom kubeconfig
bash k8s_inspect.sh --kubeconfig /path/to/kubeconfig

# Custom output directory
bash k8s_inspect.sh --kubeconfig /path/to/kubeconfig --output-dir /tmp/my-reports

Essential Operations

Run Full Inspection

# The script outputs the report file path as the last line of stdout
REPORT_PATH=$(bash k8s_inspect.sh --kubeconfig /path/to/kubeconfig 2>/dev/null | tail -1)
echo "Report generated at: $REPORT_PATH"

Process the Output

The script generates an HTML file and prints its path. Agents should:
1. Run the script and capture the output path
2. Read the HTML file or serve it for the user to view
3. Optionally parse the structured summary printed to stdout

Output Format

The script prints a structured summary before the file path:

===== INSPECTION SUMMARY =====
Nodes: 3 total, 3 ready
Pods: 45 total, 42 running, 2 abnormal, 1 pending
Warning Events: 3
Report: /tmp/k8s-inspect/k8s_report_20260417_120000.html
===============================

Inspection Modules

Common Workflows

1. Daily Health Check

Schedule via cron or K8s CronJob to generate daily reports.

2. Incident Triage

When an alert fires, run immediately to capture the cluster state snapshot.

3. Pre-deployment Audit

Run before major deployments to ensure cluster is healthy.

Security Notes

The script only performs read-only operations (kubectl get, kubectl top, kubectl describe). No cluster state is modified.

Environment Variables

• KUBECONFIG — Path to kubeconfig file (alternative to --kubeconfig flag)
• REPORT_DIR — Output directory for reports (alternative to --output-dir flag)

Differences from k8s-report-skills (Python)

Version: 1.0.0
License: MIT
Compatible with: kubectl v1.20+, Kubernetes v1.20+, Bash 4+