k8s

Kubernetes cluster management.

The k8s command provides a comprehensive interface for managing Kubernetes clusters and workloads. It supports creating clusters on EKS, GKE, and kubeadm, deploying applications, viewing resources and metrics, tailing logs, and asking natural language questions about your cluster.

Usage

clanker k8s <subcommand> [flags]

Subcommands

Subcommand	Description
create eks	Create an Amazon EKS cluster
create kubeadm	Create a kubeadm cluster on EC2 instances
create gke	Create a Google Kubernetes Engine (GKE) cluster
deploy	Deploy an application to the cluster
resources	Fetch all Kubernetes resources for visualization
logs	Get logs from a pod
stats nodes	Get node metrics
stats pods	Get pod metrics
stats pod	Get metrics for a specific pod
stats cluster	Get cluster-wide metrics
list	List Kubernetes clusters
delete	Delete a Kubernetes cluster
kubeconfig	Get kubeconfig for a cluster
ask	Ask natural language questions about your cluster

---

create eks

Create a new Amazon EKS cluster.

Usage

clanker k8s create eks [cluster-name] [flags]

Flags

Flag	Type	Default	Description
--nodes	int	1	Number of worker nodes
--node-type	string	"t3.small"	EC2 instance type for worker nodes
--version	string	"1.29"	Kubernetes version
--plan	bool	false	Show the execution plan without applying
--apply	bool	false	Apply the plan without prompting for confirmation

Examples

# Create a basic EKS cluster (prompts for confirmation)
clanker k8s create eks my-cluster

# Create with 3 worker nodes
clanker k8s create eks my-cluster --nodes 3

# Create with a specific instance type and version
clanker k8s create eks my-cluster --nodes 2 --node-type t3.medium --version 1.29

# Preview the plan without creating anything
clanker k8s create eks my-cluster --nodes 2 --plan

# Create without confirmation prompt
clanker k8s create eks my-cluster --nodes 2 --apply

---

create kubeadm

Create a kubeadm-based Kubernetes cluster on EC2 instances.

This creates a self-managed Kubernetes cluster using kubeadm. Clanker provisions EC2 instances for the control plane and workers, configures SSH access, installs Kubernetes components, and initializes the cluster with a CNI plugin.

Usage

clanker k8s create kubeadm [cluster-name] [flags]

Flags

Flag	Type	Default	Description
--workers	int	1	Number of worker nodes
--node-type	string	"t3.small"	EC2 instance type for all nodes (control plane and workers)
--key-pair	string	""	AWS key pair name for SSH access. Auto-creates as clanker-<name>-key if not specified.
--ssh-key	string	""	Path to SSH private key. Defaults to ~/.ssh/<key-pair>
--version	string	"1.29"	Kubernetes version
--plan	bool	false	Show the execution plan without applying
--apply	bool	false	Apply the plan without prompting for confirmation

Examples

# Create a basic kubeadm cluster
clanker k8s create kubeadm my-cluster

# Create with 2 workers and a specific key pair
clanker k8s create kubeadm my-cluster --workers 2 --key-pair my-key

# Create with a larger instance type
clanker k8s create kubeadm my-cluster --workers 3 --node-type t3.medium

# Specify a custom SSH key path
clanker k8s create kubeadm my-cluster --key-pair my-key --ssh-key ~/.ssh/my-key.pem

# Preview the plan
clanker k8s create kubeadm my-cluster --workers 2 --plan

# Create without confirmation
clanker k8s create kubeadm my-cluster --workers 2 --apply

---

create gke

Create a new Google Kubernetes Engine (GKE) cluster.

Usage

clanker k8s create gke [cluster-name] [flags]

Flags

Flag	Type	Default	Description
--gcp-project	string	required	GCP project ID
--gcp-region	string	"us-central1"	GCP region for the cluster
--nodes	int	1	Number of worker nodes
--node-type	string	"e2-standard-2"	GCE machine type for worker nodes
--version	string	""	Kubernetes version (uses GKE default if empty)
--preemptible	bool	false	Use preemptible (spot) VMs for worker nodes to reduce cost
--plan	bool	false	Show the execution plan without applying
--apply	bool	false	Apply the plan without prompting for confirmation

Examples

# Create a basic GKE cluster
clanker k8s create gke my-cluster --gcp-project my-project

# Create in a specific region with multiple nodes
clanker k8s create gke my-cluster --gcp-project my-project --gcp-region europe-west1 --nodes 3

# Create with a custom machine type
clanker k8s create gke my-cluster --gcp-project my-project --node-type e2-standard-4

# Use preemptible VMs for cost savings
clanker k8s create gke my-cluster --gcp-project my-project --nodes 3 --preemptible

# Preview the plan
clanker k8s create gke my-cluster --gcp-project my-project --plan

# Create without confirmation
clanker k8s create gke my-cluster --gcp-project my-project --apply

---

deploy

Deploy an application (container image) to the current Kubernetes cluster.

This command generates a Kubernetes Deployment and a LoadBalancer Service, then applies them to the cluster.

Usage

clanker k8s deploy [image] [flags]

Flags

Flag	Type	Default	Description
--name	string	""	Deployment name. Defaults to the image name (without registry prefix or tag).
--port	int	80	Container port to expose
--replicas	int	1	Number of pod replicas
--namespace	string	"default"	Kubernetes namespace to deploy into
--plan	bool	false	Show the execution plan without applying
--apply	bool	false	Apply the plan without prompting for confirmation

Examples

# Deploy nginx (prompts for confirmation)
clanker k8s deploy nginx

# Deploy with a custom name and port
clanker k8s deploy nginx --name my-web --port 8080

# Deploy with multiple replicas
clanker k8s deploy nginx --name my-web --port 80 --replicas 3

# Deploy to a specific namespace
clanker k8s deploy my-registry/my-app:latest --namespace staging

# Preview the plan
clanker k8s deploy nginx --plan

# Deploy without confirmation
clanker k8s deploy nginx --name my-web --port 80 --apply

# Deploy a private registry image
clanker k8s deploy 123456789012.dkr.ecr.us-east-1.amazonaws.com/my-app:v1.2.3 --port 3000 --apply

---

resources

Fetch all Kubernetes resources from one or more clusters for visualization.

When a cluster name is specified, it fetches resources from that cluster. When no cluster is specified, it iterates over all EKS clusters in your account and fetches resources from each.

Usage

clanker k8s resources [flags]

Flags

Flag	Type	Default	Description
--cluster	string	""	Cluster name. Uses the current kubectl context if not specified.
-o, --output	string	"json"	Output format: json or yaml

Examples

# Get resources from the current cluster context
clanker k8s resources

# Get resources from a specific EKS cluster
clanker k8s resources --cluster my-cluster

# Output as YAML
clanker k8s resources --cluster my-cluster -o yaml

# Get resources from all EKS clusters
clanker k8s resources

---

logs

Get logs from a Kubernetes pod.

Usage

clanker k8s logs [pod-name] [flags]

Flags

Flag	Type	Default	Description
-c, --container	string	""	Container name (required for multi-container pods)
-f, --follow	bool	false	Stream log output in real time
-p, --previous	bool	false	Show logs from the previous terminated container instance
--tail	int	100	Number of lines to show from the end of the logs
--since	string	""	Show logs since a relative duration (for example, 1h, 30m, 5s)
--timestamps	bool	false	Include timestamps on each log line
--all-containers	bool	false	Show logs from all containers in the pod
-n, --namespace	string	"default"	Kubernetes namespace

Examples

# Get the last 100 lines of logs from a pod
clanker k8s logs my-pod

# Stream logs in real time
clanker k8s logs my-pod -f

# Get logs from a specific container in a multi-container pod
clanker k8s logs my-pod -c sidecar

# Get logs from the previously terminated container
clanker k8s logs my-pod -p

# Get the last 500 lines
clanker k8s logs my-pod --tail 500

# Get logs from the last hour
clanker k8s logs my-pod --since 1h

# Get logs with timestamps
clanker k8s logs my-pod --timestamps

# Get logs from all containers
clanker k8s logs my-pod --all-containers

# Get logs from a pod in a specific namespace
clanker k8s logs my-pod -n kube-system

# Combine flags: stream logs from the last 30 minutes with timestamps
clanker k8s logs my-pod -f --since 30m --timestamps

---

stats nodes

Get CPU and memory metrics for all cluster nodes.

Requires the Kubernetes Metrics Server to be installed in the cluster.

Usage

clanker k8s stats nodes [flags]

Flags

Flag	Type	Default	Description
--sort-by	string	""	Sort results by cpu or memory
-o, --output	string	"table"	Output format: table, json, or yaml

Examples

# Show node metrics in table format
clanker k8s stats nodes

# Sort by CPU usage
clanker k8s stats nodes --sort-by cpu

# Sort by memory usage
clanker k8s stats nodes --sort-by memory

# Output as JSON
clanker k8s stats nodes -o json

# Output as YAML
clanker k8s stats nodes -o yaml

---

stats pods

Get CPU and memory metrics for pods.

Requires the Kubernetes Metrics Server to be installed in the cluster.

Usage

clanker k8s stats pods [flags]

Flags

Flag	Type	Default	Description
-n, --namespace	string	"default"	Kubernetes namespace
-A, --all-namespaces	bool	false	Show pods from all namespaces
--sort-by	string	""	Sort results by cpu or memory
-o, --output	string	"table"	Output format: table, json, or yaml

Examples

# Show pod metrics in the default namespace
clanker k8s stats pods

# Show pod metrics in a specific namespace
clanker k8s stats pods -n kube-system

# Show pod metrics across all namespaces
clanker k8s stats pods -A

# Sort by memory usage
clanker k8s stats pods --sort-by memory

# Output as JSON
clanker k8s stats pods -A -o json

---

stats pod

Get CPU and memory metrics for a specific pod and its containers.

Requires the Kubernetes Metrics Server to be installed in the cluster.

Usage

clanker k8s stats pod [pod-name] [flags]

Flags

Flag	Type	Default	Description
-n, --namespace	string	"default"	Kubernetes namespace
--containers	bool	false	Show per-container metrics breakdown
-o, --output	string	"table"	Output format: table, json, or yaml

Examples

# Show metrics for a specific pod
clanker k8s stats pod my-pod

# Show per-container metrics
clanker k8s stats pod my-pod --containers

# Show metrics for a pod in a specific namespace
clanker k8s stats pod my-pod -n kube-system

# Output as JSON
clanker k8s stats pod my-pod --containers -o json

---

stats cluster

Get aggregated CPU and memory metrics for the entire cluster.

This command summarizes node-level metrics into cluster-wide averages.

Usage

clanker k8s stats cluster [flags]

Flags

Flag	Type	Default	Description
-o, --output	string	"table"	Output format: table, json, or yaml

Examples

# Show cluster-wide metrics summary
clanker k8s stats cluster

# Output as JSON
clanker k8s stats cluster -o json

# Output as YAML
clanker k8s stats cluster -o yaml

---

list

List Kubernetes clusters of a specific type.

Usage

clanker k8s list [cluster-type] [flags]

The cluster-type argument defaults to eks. Supported types: eks, gke, kubeadm.

Flags

Flag	Type	Default	Description
--gcp-project	string	""	GCP project ID (required for gke type)
--gcp-region	string	""	GCP region (used with gke type)

Examples

# List EKS clusters (default)
clanker k8s list

# List EKS clusters explicitly
clanker k8s list eks

# List GKE clusters
clanker k8s list gke --gcp-project my-project

# List GKE clusters in a specific region
clanker k8s list gke --gcp-project my-project --gcp-region europe-west1

# List kubeadm clusters
clanker k8s list kubeadm

---

delete

Delete an existing Kubernetes cluster.

This command displays a plan of what will be deleted and prompts for confirmation before proceeding.

Usage

clanker k8s delete [cluster-type] [cluster-name] [flags]

Supported cluster types: eks, gke, kubeadm.

Flags

Flag	Type	Default	Description
--gcp-project	string	""	GCP project ID (required for gke type)
--gcp-region	string	""	GCP region (used with gke type)

Examples

# Delete an EKS cluster (prompts for confirmation)
clanker k8s delete eks my-cluster

# Delete a GKE cluster
clanker k8s delete gke my-cluster --gcp-project my-project

# Delete a GKE cluster in a specific region
clanker k8s delete gke my-cluster --gcp-project my-project --gcp-region us-west1

# Delete a kubeadm cluster
clanker k8s delete kubeadm my-cluster

---

kubeconfig

Retrieve and configure kubeconfig for a cluster.

After running this command, you can use kubectl to interact with the cluster.

Usage

clanker k8s kubeconfig [cluster-type] [cluster-name] [flags]

Supported cluster types: eks, gke, kubeadm.

Flags

Flag	Type	Default	Description
--gcp-project	string	""	GCP project ID (required for gke type)
--gcp-region	string	""	GCP region (used with gke type)

Examples

# Get kubeconfig for an EKS cluster
clanker k8s kubeconfig eks my-cluster

# Get kubeconfig for a GKE cluster
clanker k8s kubeconfig gke my-cluster --gcp-project my-project

# Get kubeconfig for a kubeadm cluster
clanker k8s kubeconfig kubeadm my-cluster

# Typical workflow after getting kubeconfig
clanker k8s kubeconfig eks my-cluster
kubectl get nodes
kubectl get pods -A

---

ask

Ask natural language questions about your Kubernetes cluster using AI.

The ask subcommand uses a three-stage LLM pipeline to analyze your question, execute the appropriate kubectl operations, and return a comprehensive response. Conversation history is maintained per cluster for follow-up questions.

Usage

clanker k8s ask [question] [flags]

Flags

Flag	Type	Default	Description
--cluster	string	""	Kubernetes cluster name (EKS or GKE cluster name). Updates kubeconfig automatically.
--profile	string	""	AWS profile for EKS clusters
--kubeconfig	string	""	Path to kubeconfig file (default: ~/.kube/config)
--context	string	""	kubectl context to use (overrides --cluster)
-n, --namespace	string	""	Default namespace for queries. Defaults to all namespaces when empty.
--ai-profile	string	""	AI profile to use for LLM queries
--debug	bool	false	Enable debug output
--gcp	bool	false	Use GKE cluster instead of EKS
--gcp-project	string	""	GCP project ID for GKE clusters
--gcp-region	string	""	GCP region for GKE clusters

Examples

# Ask about pod status in the current cluster context
clanker k8s ask "how many pods are running"

# Target a specific EKS cluster
clanker k8s ask --cluster my-cluster "show me all deployments"

# Target an EKS cluster with a specific AWS profile
clanker k8s ask --cluster prod --profile production "give me error logs for nginx pod"

# Target a GKE cluster
clanker k8s ask --gcp --gcp-project my-project --cluster my-gke-cluster "show me all pods"

# Ask about resource usage
clanker k8s ask "which pods are using the most memory"

# Troubleshoot issues
clanker k8s ask "why is my pod crashing"

# Get a cluster health overview
clanker k8s ask "tell me the health of my cluster"

# Use a specific namespace context
clanker k8s ask -n kube-system "show me all services"

# Use a specific kubeconfig file
clanker k8s ask --kubeconfig /path/to/kubeconfig "list all nodes"

# Use a specific kubectl context
clanker k8s ask --context arn:aws:eks:us-east-1:123456789012:cluster/my-cluster "show deployments"

# Use a specific AI provider
clanker k8s ask --ai-profile anthropic "analyze my cluster security"

# Enable debug output to see the LLM pipeline stages
clanker k8s ask --debug "show me pods with restart counts"

How It Works

The k8s ask command follows a three-stage pipeline:

1. Stage 1: Query Analysis -- The LLM analyzes your question along with the current cluster status (node count, pod counts, namespaces) and determines which kubectl operations are needed.

2. Stage 2: Execution -- The identified kubectl operations are executed against the cluster (for example, kubectl get pods -A, kubectl describe pod my-pod, kubectl top nodes).

3. Stage 3: Response -- The kubectl output is combined with cluster context and conversation history, then the LLM generates a comprehensive markdown-formatted response.

Conversation history is persisted per cluster, so you can ask follow-up questions without restating context.

---

Prerequisites

Different subcommands require different tools to be installed:

Tool	Required For
aws CLI	EKS cluster operations, kubeadm on EC2
gcloud CLI	GKE cluster operations
kubectl	All cluster interaction (deploy, logs, stats, ask, resources)
eksctl	EKS cluster creation
helm	Helm-based deployments
Metrics Server	stats subcommands (must be installed in the cluster)

See Also

• ask -- The primary query command with K8s auto-routing
• deploy -- Deploy repositories to the cloud
• AWS Cloud Provider -- AWS-specific configuration
• GCP Cloud Provider -- GCP-specific configuration