Back to list
eveld

k8s-awareness

by eveld

Research, planning, and implementation workflows for Claude Code

3🍴 1📅 Jan 6, 2026

SKILL.md


Kubernetes Skills Guide

You have specialized Kubernetes debugging skills. Use these instead of raw kubectl commands for consistent, well-documented workflows.

Decision Tree

Skills vs Agents

Simple, single kubectl command query? → Use kubernetes skill

  • Better than: Running raw kubectl get pods commands
  • Example: "Check pod status in production namespace"
  • Use when: Quick resource check, single namespace, no correlation needed

Complex investigation requiring multiple steps? → Use Task tool with K8s agents (conserves context)

  • k8s-locator - Find and list resources across namespaces
  • k8s-analyzer - Diagnose pod/deployment issues, check logs/events
  • k8s-pattern-finder - Find patterns across resources, detect infrastructure issues
  • Use when: Multi-namespace investigation, pattern detection, cluster-wide issues

Specific pod crashing or failing? → Use k8s-debug skill

  • Launches ephemeral debug container in running pod
  • Example: "Debug the failing pod api-gateway-xyz"

When to Use Which Agent

Just need to find resources broadly? → Use k8s-locator agent only

  • Lists pods, deployments, services across namespaces
  • Saves to /tmp for later analysis
  • Example: "Get all pods in production, staging, development namespaces"

Investigating single pod/service issue? → Use k8s-locator + k8s-analyzer agents

  • Locator: Find relevant pods/resources
  • Analyzer: Diagnose health, check events, analyze logs
  • Example: "Debug service-b CrashLoopBackOff"

Need to find patterns or cluster-wide issues? → Use all three: k8s-locator + k8s-analyzer + k8s-pattern-finder

  • Locator: Fetch resources from all namespaces
  • Analyzer: Diagnose specific pod failures
  • Pattern-finder: Correlate across resources, detect node issues, cascade failures
  • Example: "Find why multiple pods are ImagePullBackOff across namespaces"

Available Kubernetes Tools

TypeNamePurpose
SkillkubernetesSimple queries (single kubectl command)
Skillk8s-debugLaunch ephemeral debug container
Agentk8s-locatorFind and list resources across namespaces
Agentk8s-analyzerDiagnose pod/deployment issues
Agentk8s-pattern-finderFind patterns, cluster-wide issues

When to Use Raw kubectl

Only use kubectl directly when:

  • Running one-off commands not covered by skills
  • User explicitly requests a specific kubectl command
  • Debugging the skill itself

For systematic Kubernetes work, use the specialized skills above.

Common kubectl Commands

Covered by skills:

  • kubectl get → Use kubernetes
  • kubectl describe → Use kubernetes
  • kubectl logs → Use kubernetes
  • kubectl debug → Use k8s-debug

Not covered yet (use directly):

  • kubectl apply, kubectl delete, kubectl edit (destructive operations)
  • kubectl port-forward, kubectl exec (interactive operations)

Authentication

Check context before any kubectl operation:

kubectl config current-context
kubectl config view --minify

Score

Total Score

60/100

Based on repository quality metrics

SKILL.md

SKILL.mdファイルが含まれている

+20
LICENSE

ライセンスが設定されている

+10
説明文

100文字以上の説明がある

0/10
人気

GitHub Stars 100以上

0/15
最近の活動

3ヶ月以内に更新

+5
フォーク

10回以上フォークされている

0/5
Issue管理

オープンIssueが50未満

+5
言語

プログラミング言語が設定されている

0/5
タグ

1つ以上のタグが設定されている

+5

Reviews

💬

Reviews coming soon