skills.rest
Aki2022
Aki2022COMMUNITY
Explore all skills by this author →
GitHub

gke-basics

Plan, create, and configure Google Kubernetes Engine clusters using golden path Autopilot defaults.

Updated Dec 9, 2025
Search Tags:#devops#kubernetes#containers#google cloud#gke#autopilot#cluster provisioning
One-click install
npx skills add https://github.com/Aki2022/skills --skill gke-basics-aki2022
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: gke-basics
Source: https://github.com/Aki2022/skills/tree/main/gke-basics
Command: npx skills add https://github.com/Aki2022/skills --skill gke-basics-aki2022

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) and assets (resource) components.

What problem does it solve? Designing and operating production GKE clusters involves dozens of decisions across networking, security, scaling, and cost, and misconfigurations made at creation time (Day-0) are hard or impossible to change later. This Skill provides a golden path Autopilot configuration with opinionated defaults, guardrails, and per-scenario reference guides so clusters are provisioned correctly the first time. ## Core Features & Use Cases - Golden Path Cluster Creation: Create production-ready Autopilot clusters with private nodes, Workload Identity, Secret Manager rotation, hardened RBAC, and full observability via MCP tools or gcloud CLI. - Scenario-Based References: Twenty-plus reference guides covering networking, security, scaling, ComputeClasses, cost optimization, AI/ML inference with GIQ, upgrades, multi-tenancy, batch/HPC, backup/DR, storage, and reliability. - Tool Preference Chain: Structured guidance on when to use GKE MCP tools, gcloud CLI, or kubectl, with operation-level coverage matrices and error remediation tables. - Use Case: A platform engineer asks to provision a new production cluster; the Skill gathers project, region, and name, applies the golden path Autopilot template with private nodes and Secret Manager, creates the cluster via MCP, and verifies the applied settings. ## Quick Start Ask the assistant to create a production GKE Autopilot cluster named my-cluster in us-central1 using the golden path configuration.

Frequently Asked Questions about gke-basics

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I create a production GKE Autopilot cluster?▼

Use gcloud container clusters create-auto with the golden path flags: --enable-private-nodes, --enable-master-authorized-networks, --enable-dns-access, --enable-secret-manager, and --release-channel regular. Alternatively, call the MCP create_cluster tool with the equivalent JSON cluster configuration.

What is the difference between GKE Autopilot and Standard mode?▼

Autopilot is Google-managed: nodes, scaling, and security are handled for you and you pay per pod resource request. Standard gives full control over node pools, OS, and machine types, and is only recommended when Autopilot has a documented limitation such as kernel customization needs.

How does Workload Identity work on GKE?▼

Workload Identity lets pods access Google Cloud APIs without static service account keys. You create a Google service account, bind it to a Kubernetes service account with the iam.workloadIdentityUser role, annotate the KSA, and the pod authenticates as the GSA.

Can I run GPU inference workloads on GKE Autopilot?▼

Yes. Autopilot supports GPUs through ComputeClasses that target machine families like g2 with NVIDIA L4. You can also use gcloud container ai profiles to generate optimized inference manifests for models like Gemma or Llama with vLLM.

Why can't I change my GKE cluster's VPC or subnet after creation?▼

VPC, subnet, and IP range allocation are Day-0 decisions that are immutable after cluster creation. Changing them requires recreating the cluster, so the Skill flags these choices prominently and asks clarifying questions before provisioning.

How do I reduce GKE costs for batch workloads?▼

Use Spot VMs via a ComputeClass with on-demand fallback and activeMigration, which can cut costs 60-90 percent for fault-tolerant jobs. Combine with VPA rightsizing recommendations, scale-to-zero node pools, and committed use discounts for steady-state usage.

Related Skills

Explore More Software Engineering Skills→
kubiyabotkubiyabot
16

gcp

Manage Google Cloud resources by wrapping the native gcloud CLI.

Official
Advanced
oliverconstanceoliverconstance

gcp-cloud-run-deploy

Deploys containerized microservices to Google Cloud Run using gcloud CLI and Knative manifests.

Community
Intermediate
wangx7wangx7

gke-golden-path

Generate production-ready GKE Autopilot cluster configurations with security and networking defaults.

Community
Advanced
rnavarychrnavarych
14

role-database:database-devops

Automate database provisioning, Kubernetes deployment, and CI/CD integration with Terraform.

Community
Advanced