gke-cluster-creator
Guides the user through creating GKE clusters using pre-defined templates (Standard, Autopilot, GPU/AI).
How do I install this agent skill?
npx skills add https://github.com/googlecloudplatform/gke-mcp --skill gke-cluster-creatorIs this agent skill safe to install?
- Gen Agent Trust Hubpass
The skill is a legitimate template-based guide for creating Google Kubernetes Engine (GKE) clusters using a specific MCP tool. It follows security best practices and contains no malicious patterns or high-risk behaviors.
- Socketpass
No alerts
- Snykpass
Risk: LOW · No issues
What does this agent skill do?
GKE Cluster Creation Skill
This skill helps users create Google Kubernetes Engine (GKE) clusters by providing a set of best-practice templates and guiding them through the customization process.
core_behavior
- Template Selection:
- Present the available templates to the user if they haven't specified one.
- Explain the trade-offs (e.g., Cost vs. Availability, Autopilot vs. Standard).
- Customization:
- Once a template is selected, present the default configuration (JSON/YAML).
- Ask the user for essential missing information:
project_id,location,cluster_name. - Ask if they want to modify optional fields (e.g.,
machineType,nodeCount,network).
- Validation:
- Ensure
project_id,location, andcluster_nameare set. - Ensure the configuration matches the
create_clusterMCP tool schema.
- Ensure
- Execution:
- Call the
create_clusterMCP tool with the final configuration.
- Call the
best_practices
When guiding the user or generating configurations, adhere to the following GKE cluster creation best practices:
Security
- Private Clusters: Default to private clusters with a private control plane and restricted public endpoints to minimize attack surface.
- VPC-Native Networking: Use VPC-native clusters to enable alias IP ranges, which allows pod-level firewall rules and better network security.
- Workload Identity: Prefer Workload Identity for securely granting GKE workloads access to Google Cloud services instead of using static service account keys.
- Shielded GKE Nodes: Enable Shielded GKE Nodes to protect against rootkits and bootkits.
- Least Privilege (RBAC): Institute strict Role-Based Access Control limits granting minimal privilege to users and workloads.
Cost Optimization
- Autoscaling: Enable Cluster Autoscaler and Horizontal Pod Autoscaler to adjust resources based on demand.
- Right-Sizing: Choose the appropriate machine types and node counts. Consider Spot VMs for fault-tolerant, non-critical workloads.
High Availability & Reliability
- Regional Clusters: Use Regional Clusters for production environments to ensure control plane replication across multiple zones. (Note: standard regional creates nodes across 3 zones by default).
- Pod Disruption Budgets: Recommend setting Pod Disruption Budgets for application stability during node maintenance.
- Release Channels: Subscribe to a release channel (e.g., Regular or Stable) for automated and safer cluster upgrades.
templates
1. Standard Zonal (Cost-Effective Dev/Test)
Best for: Development, testing, non-critical workloads.
{
"name": "projects/{PROJECT_ID}/locations/{ZONE}/clusters/{CLUSTER_NAME}",
"initialNodeCount": 1,
"nodeConfig": {
"machineType": "e2-medium",
"diskSizeGb": 50,
"oauthScopes": [
"https://www.googleapis.com/auth/devstorage.read_only",
"https://www.googleapis.com/auth/logging.write",
"https://www.googleapis.com/auth/monitoring",
"https://www.googleapis.com/auth/service.management.readonly",
"https://www.googleapis.com/auth/servicecontrol",
"https://www.googleapis.com/auth/trace.append"
]
}
}
2. Standard Regional (High Availability)
Best for: Production workloads requiring high availability. Note: Creates 3 nodes (one per zone in the region) by default.
{
"name": "projects/{PROJECT_ID}/locations/{REGION}/clusters/{CLUSTER_NAME}",
"initialNodeCount": 1,
"nodeConfig": {
"machineType": "e2-standard-4",
"diskSizeGb": 100,
"oauthScopes": ["https://www.googleapis.com/auth/cloud-platform"]
}
}
3. Autopilot (Operations-Free)
Best for: Most workloads where you don't want to manage nodes.
{
"name": "projects/{PROJECT_ID}/locations/{REGION}/clusters/{CLUSTER_NAME}",
"autopilot": {
"enabled": true
}
}
4. GPU Inference (L4)
Best for: AI/ML Inference, small model serving.
Note: Requires g2-standard-4 quota.
{
"name": "projects/{PROJECT_ID}/locations/{REGION}/clusters/{CLUSTER_NAME}",
"initialNodeCount": 1,
"nodeConfig": {
"machineType": "g2-standard-4",
"accelerators": [
{
"acceleratorCount": "1",
"acceleratorType": "nvidia-l4"
}
],
"diskSizeGb": 100,
"oauthScopes": ["https://www.googleapis.com/auth/cloud-platform"]
}
}
5. AI Hypercompute (A3 HighGPU)
Best for: Large Model Training/Inference. Note: High cost and strict quota requirements.
{
"name": "projects/{PROJECT_ID}/locations/{REGION}/clusters/{CLUSTER_NAME}",
"initialNodeCount": 1,
"nodeConfig": {
"machineType": "a3-highgpu-8g",
"accelerators": [
{
"acceleratorCount": "8",
"acceleratorType": "nvidia-h100-80gb-hbm3"
}
],
"diskSizeGb": 200,
"oauthScopes": ["https://www.googleapis.com/auth/cloud-platform"]
}
}
instructions
- ALWAYS ask for the
project_idif it is not in the context. - ALWAYS ask for the
location(Region or Zone). - ALWAYS ask for a unique
cluster_name. - CHECK if the user wants
Access to Google Cloud APIs(defaultcloud-platformscope is usually best for modern GKE). - WARN the user about cost if they select GPU or Reginal clusters.
- USE
create_clusterMCP tool to create the cluster. Theparentargument isprojects/{PROJECT_ID}/locations/{LOCATION}and theclusterargument is the JSON object. Thecluster.nameis just the short name (e.g. "my-cluster"). - IMPORTANT: When calling
create_cluster, thecluster.nameshould be the short name (e.g.,my-cluster), NOT the full resource path, because theparentargument defines the scope.
example_usage
User: "I want to create a GKE cluster." Model: "I can help with that. What kind of cluster do you need?
- Standard Zonal: Good for dev/test.
- Standard Regional: High availability.
- Autopilot: Fully managed.
- GPU Enabled: For AI/ML workloads."
User: "Standard Zonal, please." Model: "Great. I'll need a few details:
- Project ID
- Zone (e.g., us-central1-a)
- Cluster Name"
User: "Project my-proj, zone us-west1-b, name dev-cluster."
Model: "Here is the configuration I will use:
[JSON view]
Do you want to proceed?"
How can the creator link this skill?
Add the canonical catalog link to the repository README so users can inspect current installs and available audits. The publishing guide covers the complete discovery path.
<a href="https://skillzs.dev/skills/googlecloudplatform/gke-mcp/gke-cluster-creator">View gke-cluster-creator on skillZs</a>