K8sGPT Cheatsheet

K8sGPT is an AI-powered diagnostic tool for Kubernetes clusters that helps identify and solve issues using natural language processing. This cheatsheet provides a quick reference for common K8sGPT commands and operations.

Want to try it now? Do check out my step by step interactive tutorial on K8sGPT here: https://killercoda.com/kubetools/scenario/k8sGPT

Installation

Method	Command
Homebrew	brew tap k8sgpt-ai/k8sgpt && brew install k8sgpt
Binary	Download from GitHub Releases
Krew	kubectl krew install k8sgpt
Go	go install github.com/k8sgpt-ai/k8sgpt@latest

Basic Commands

Operation	Command	Description
Version	k8sgpt version	Display the current version of K8sGPT
Help	k8sgpt --help	Show help information
Authentication	k8sgpt auth	Configure authentication for AI providers
Analyze	k8sgpt analyze	Run analysis on your Kubernetes cluster
List	k8sgpt list	List all analyses that have been performed
Serve	k8sgpt serve	Start K8sGPT in server mode
Filters	k8sgpt filters	Show and manage resource filters

Configuration

Operation	Command	Description
Configure Provider	k8sgpt auth --provider <provider>	Set the AI provider (openai, azure, etc.)
Set API key	k8sgpt auth --apikey <your-api-key>	Configure your AI provider API key
Set Model	k8sgpt auth --model <model-name>	Set the AI model to use
Configure Context	k8sgpt auth --context <context-name>	Set the Kubernetes context

Analysis Options

Operation	Command	Description
Basic Analysis	k8sgpt analyze	Run analysis on all resources
Filter by Namespace	k8sgpt analyze --namespace <ns>	Analyze resources in specific namespace
Filter by Kind	k8sgpt analyze --filter <kind>	Analyze only specific resource types
Specify Kubeconfig	k8sgpt analyze --kubeconfig <path>	Use specific kubeconfig file
Set Severity	k8sgpt analyze --threshold <num>	Filter by minimum severity (0-10)
Generate YAML	k8sgpt analyze --explain yaml	Generate explanation in YAML format
Generate JSON	k8sgpt analyze --explain json	Generate explanation in JSON format
Cache Results	k8sgpt analyze --cache	Enable caching of results
Max Results	k8sgpt analyze --max-results <num>	Limit number of results

Filter Management

Operation	Command	Description
List Filters	k8sgpt filters list	Show all active filters
Add Filter	k8sgpt filters add <kind>	Add resource kind to filter list
Remove Filter	k8sgpt filters remove <kind>	Remove resource kind from filter list
Reset Filters	k8sgpt filters reset	Reset all filters to default

Integration Options

Integration	Command/Reference	Description
Kubectl Plugin	kubectl k8sgpt analyze	Use as kubectl plugin
Kubernetes Operator	k8sgpt-operator	Deploy as an operator
Prometheus Integration	k8sgpt serve --metrics	Expose Prometheus metrics
Slack Integration	Configure via .k8sgpt.yaml	Send results to Slack

Example Workflows

Task	Command Sequence	Description
Quick Troubleshooting	k8sgpt analyze	Run a full cluster analysis
Focused Analysis	k8sgpt analyze --filter Pod --namespace app	Analyze only Pods in the app namespace
Continuous Monitoring	k8sgpt serve	Run in server mode for continuous analysis
CI/CD Integration	k8sgpt analyze --explain json --output file.json	Generate machine-readable output for CI/CD

Advanced Configuration (.k8sgpt.yaml)

ai:
  provider: openai
  model: gpt-4
  apikey: your-api-key-here
  backend: http://localhost:8080
  temperature: 0.7
filters:
  - Pod
  - Deployment
  - StatefulSet
kubernetes:
  context: my-cluster-context
  kubeconfig: /path/to/kubeconfig

Supported Resource Kinds

Category	Resource Kinds
Workloads	Pod, Deployment, StatefulSet, DaemonSet, ReplicaSet, Job, CronJob
Services	Service, Ingress, NetworkPolicy
Config	ConfigMap, Secret, HorizontalPodAutoscaler
Storage	PersistentVolume, PersistentVolumeClaim, StorageClass
RBAC	Role, RoleBinding, ClusterRole, ClusterRoleBinding, ServiceAccount
Custom	CustomResourceDefinition, plus most custom resources

Tips and Best Practices

• Start with a broad analysis and then narrow down with filters
• Use the --explain yaml option for sharing analysis results with team members
• Configure multiple AI providers as fallbacks
• Set appropriate severity thresholds for your environment
• Integrate with monitoring tools for continuous analysis