Skip to content

cloudthinker-ai/CloudSkills / skills/connections/k8s

MANDATORY parallel execution patterns, cluster overview script, -o json with jq filtering, batch operations, and common pitfalls

Kubernetes CLI Skill

Execute kubectl and helm commands with proper kubeconfig injection.

Workflow

bun run ./_skills/connections/k8s/k8s/scripts/discover.ts

The output is auto-cached. Do not re-run unless user explicitly requests refresh.

CLI Tips

Recommended First Step

Run the cluster discovery script (or the shell script directly):

bun run ./_skills/connections/k8s/k8s/scripts/discover.ts
# or
./get_k8s_cluster_overview.sh

This provides comprehensive cluster analysis (nodes, pods, resources, services, events) in a single parallelized call. Only run targeted kubectl commands AFTER reviewing the overview output for specific deep-dives.

Critical Requirements

🚨 MANDATORY: ALL independent kubectl operations MUST run in parallel using background jobs (&) + wait

  • FORBIDDEN: Sequential loops for item in $items; do kubectl get $item; done
  • REQUIRED: Background jobs for item in $items; do kubectl get $item & done; wait
  • Read-only commands only (get, describe, top, logs) - never modify resources

Parallel Execution

When to parallelize:

  • Multiple namespaces/nodes/pods/resources β†’ ALWAYS parallel
  • Operations with data dependencies β†’ Sequential only

Pattern:

# CORRECT: Parallel (0.5s for 30 pods)
for pod in $pods; do kubectl get pod "$pod" -n "$ns" & done; wait

# WRONG: Sequential (15s for 30 pods)
for pod in $pods; do kubectl get pod "$pod" -n "$ns"; done

Best practice - Batch:

# Fastest: Single kubectl call per namespace
kubectl get pods -n "$ns" -o json | jq '.items[] | {name, phase, ready}'

Output Format

  • Complex data: -o json with jq (better AI parsing) kubectl get pods -n $ns -o json | jq -r '.items[] | "\(.metadata.name)\t\(.status.phase)"'
  • Simple lists: -o jsonpath or --no-headers (token efficient) kubectl get pods -n $ns -o jsonpath='{.items[*].metadata.name}'
  • CRITICAL: Always filter JSON with jq - never dump raw output
  • Machine-readable only: no decorative formatting, never expose credentials

Optimization Rules

  • Batch operations: Single kubectl get -o json instead of multiple calls BAD: kubectl get node $n -o jsonpath='{.status.nodeInfo.machineID}'; kubectl get node $n -o jsonpath='{.status.nodeInfo.kubeletVersion}' GOOD: kubectl get node $n -o json | jq '{machineID: .status.nodeInfo.machineID, kubeletVersion: .status.nodeInfo.kubeletVersion}'
  • Cache namespaces: Get once and reuse
  • Filter at source: Use --field-selector and --selector before post-processing
  • Resource usage: kubectl top requires metrics-server; handle gracefully if unavailable

Common Patterns

  • List: kubectl get <resource> -n <ns> -o json | jq -r '.items[] | ...'
  • Filter by label: kubectl get pods -n <ns> -l app=myapp
  • Filter by field: kubectl get pods --field-selector=status.phase=Running
  • Events: kubectl get events -n <ns> --sort-by='.lastTimestamp'
  • Logs: kubectl logs <pod> -n <ns> --tail=100 (use cautiously)

Common Pitfalls

  • Multiple calls for same resource β†’ Use single -o json + jq
  • Not caching namespace list β†’ Cache once
  • Sequential namespace processing β†’ Always parallelize
  • Raw JSON dumps β†’ Always filter with jq first
  • kubectl describe verbose β†’ Use kubectl get -o json | jq instead

Helper Script

./get_k8s_cluster_overview.sh (in sandbox home) - YOUR GO-TO STARTING POINT Provides: cluster info, nodes, pods, resources, namespaces, services, events Usage: ./get_k8s_cluster_overview.sh Output: Structured JSON or machine-readable text (all operations parallelized) TIP: Run this FIRST, then drill down into specific areas based on findings

Anti-Hallucination Rules

  1. NEVER assume resource names β€” always discover via CLI/API in Phase 1 before referencing in Phase 2.
  2. NEVER fabricate metric names or dimensions β€” verify against the service documentation or --help output.
  3. NEVER mix CLI commands between service versions β€” confirm which version/API you are targeting.
  4. ALWAYS use the discovery β†’ verify β†’ analyze chain β€” every resource referenced must have been discovered first.
  5. ALWAYS handle empty results gracefully β€” an empty response is valid data, not an error to retry.

Counter-Rationalizations

ShortcutCounterWhy
"I'll skip discovery and check known resources"Always run Phase 1 discovery firstResource names change, new resources appear β€” assumed names cause errors
"The user only asked for a quick check"Follow the full discovery β†’ analysis flowQuick checks miss critical issues; structured analysis catches silent failures
"Default configuration is probably fine"Audit configuration explicitlyDefaults often leave logging, security, and optimization features disabled
"Metrics aren't needed for this"Always check relevant metrics when availableAPI/CLI responses show current state; metrics reveal trends and intermittent issues
"I don't have access to that"Try the command and report the actual errorAssumed permission failures prevent useful investigation; actual errors are informative
writing-skillsobra/superpowersUse when creating new skills, editing existing skills, or verifying skills work before deploymentStars: 318kUpdated Claude CodeCodexUnverifiedInstall
xlsxanthropics/skillsUse this skill any time a spreadsheet file is the primary input or output. This means any task where the user wants to: open, read, edit, or fix an existing .xlsx, .xlsm, .xltx, .csv, or .tsv file (e.g., adding columns, computing formulas, formatting, charting, cleaning messy data); create a new spreadsheet from scratch or from other data sources; or convert between tabular file formats. Trigger especially when the user references a spreadsheet file by name or path β€” even casually (like "the xlsx in my downloads") β€” and wants something done to it or produced from it. Also trigger for cleaning or restructuring messy tabular data files (malformed rows, misplaced headers, junk data) into proper spreadsheets. The deliverable must be a spreadsheet file. Do NOT trigger when the primary deliverable is a Word document, HTML report, standalone Python script, database pipeline, or Google Sheets API integration, even if tabular data is involved.Stars: 196kUpdated Claude CodeCodexVerifiedInstall
shipping-and-launchaddyosmani/agent-skillsPrepares production launches. Use when preparing to deploy to production, or when asking what needs to be in place before shipping. Use when you need a pre-launch checklist, when setting up monitoring, when planning a staged rollout, or when you need a rollback strategy.Stars: 118kUpdated Claude CodeCodexVerifiedInstall
childhoodthedotmack/claude-memBabysit a shipped release after merge, publication, or deployment. Verify published packages, production behavior, installed plugins, and late CI or user reports until the observation window is clean; use babysit for PR readiness before release.Stars: 117kUpdated Claude CodeCodexUnverifiedInstall