Claude Plugin — Kubernetes

An MCP server that gives Claude deep Kubernetes cluster awareness, diagnostics, and remediation capabilities via kubectl.

Tool	Description
`k8s_cluster_info`	Current context, server version, API endpoint
`k8s_get_contexts`	List all kubeconfig contexts, indicate active one
`k8s_list_namespaces`	List namespaces with status and age
`k8s_list_nodes`	Nodes with roles, status, version, OS, IP, age
`k8s_list_pods`	Pods with status, restarts, node (filter by ns/label)
`k8s_list_deployments`	Deployments with replica counts and age
`k8s_list_services`	Services with type, cluster IP, external IP, ports
`k8s_list_images`	Container images running across pods
`k8s_list_events`	Cluster events sorted by time (filterable by Warning type)
`k8s_list_statefulsets`	StatefulSets with desired/ready counts and age
`k8s_list_ingresses`	Ingress resources with hosts, paths, backends
`k8s_list_jobs`	Jobs with completions, duration, age
`k8s_list_cronjobs`	CronJobs with schedule, last schedule time, active count
`k8s_list_configmaps`	ConfigMaps (metadata only)
`k8s_list_secrets`	Secrets (metadata only — never exposes data)
`k8s_list_pvcs`	PVCs with status, volume, capacity, access modes
`k8s_list_daemonsets`	DaemonSets with desired/current/ready counts
`k8s_list_hpa`	HPA with target, min/max replicas, metrics
`k8s_list_networkpolicies`	NetworkPolicies with pod selector and age
`k8s_api_resources`	Available API resource types including CRDs

Tool	Description
`k8s_describe`	`kubectl describe` any resource
`k8s_logs`	Pod logs with tail, container, previous, since, and error filtering
`k8s_top_pods`	Pod CPU/memory usage (requires metrics-server)
`k8s_top_nodes`	Node CPU/memory usage (requires metrics-server)
`k8s_find_issues`	Comprehensive health scan — pods, nodes, deployments, statefulsets, jobs, PVCs, events
`k8s_get_yaml`	Resource as YAML (managed fields stripped by default; `raw=true` for full output)
`k8s_self_test`	Plugin health check — kubectl binary, cluster connectivity, auth, metrics-server

Tool	Risk	Description
`k8s_restart_deployment`	Low	Rolling restart a deployment
`k8s_scale`	Medium	Scale deployment/statefulset (scale-to-zero requires explicit confirmation)
`k8s_delete_pod`	Low-Med	Delete pod for controller recreation; `force=true` for stuck pods
`k8s_rollback_deployment`	Medium	Roll back to previous or specific revision
`k8s_apply_manifest`	Medium	Apply YAML/JSON manifest; dry-run supported; cluster-scoped resources blocked by default
`k8s_patch_resource`	Medium	Strategic merge, JSON merge, or JSON patch any resource
`k8s_node_operation`	High	Cordon / uncordon / drain a node
`k8s_rollout_status`	Read-only	Check rollout progress (30s timeout)
`k8s_rollout_history`	Read-only	View all recorded revisions for a deployment
`k8s_delete_resource`	Medium	Delete any resource by type and name
`k8s_diff`	Read-only	Diff a manifest against live cluster state

Static resources and namespace-scoped templates for direct data access:

URI	Description
`k8s://contexts`	List kubeconfig contexts
`k8s://cluster-info`	Current context, server version, API endpoint
`k8s://namespaces`	All namespaces with status and age
`k8s://namespaces/{namespace}/pods`	Pods in a namespace
`k8s://namespaces/{namespace}/deployments`	Deployments in a namespace
`k8s://namespaces/{namespace}/services`	Services in a namespace
`k8s://namespaces/{namespace}/events`	Events in a namespace

Structured troubleshooting workflows:

Prompt	Description
`diagnose-pod`	Step-by-step pod diagnosis (describe, events, logs, resources)
`cluster-health-report`	Full health scan with drill-down into issues
`incident-response`	Incident triage — scan, events, failing pods, blast radius
`debug-crashloop`	CrashLoopBackOff debugging — previous logs, OOM check, probes
`pre-deploy-checklist`	Pre-deployment verification — deployments, pods, resources, nodes

Add to your project's .mcp.json:

{
  "mcpServers": {
    "kubernetes": {
      "command": "uvx",
      "args": ["--from", "claude-plugin-kubernetes", "k8s-mcp"]
    }
  }
}

To verify, run /mcp inside Claude Code. You should see kubernetes listed with 38 tools.

Add to ~/Library/Application Support/Claude/claude_desktop_config.json (macOS) or %APPDATA%\Claude\claude_desktop_config.json (Windows):

kubernetes