Return cluster version, node count, API server URL, and overall health status.

List all namespaces with status, age, and labels.

List cluster nodes with roles, status, capacity, and allocatable resources.

Structured node conditions (Ready, memory/disk/PID/network pressure) with unhealthy nodes first and short remediation hints. Read-only.

List pods in a namespace with status, restarts, age, and container images.

List deployments in a namespace with replica counts, strategy, and age.

List services in a namespace with type, cluster IP, ports, and selectors.

List ingress resources in a namespace with hosts, paths, and TLS config.

If gateway.networking.k8s.io is installed, list Gateways, HTTPRoutes, and GRPCRoutes. Omit namespace for all namespaces. Read-only.

List HorizontalPodAutoscalers with scaling target, min/max/current replicas, metrics, and current utilization.

List ResourceQuotas showing used vs hard limits, and LimitRanges showing min/max/default resource constraints per namespace.

List PDBs with minAvailable/maxUnavailable and current vs desired healthy pods. Omit namespace for all namespaces. Read-only.

Read-only RBAC review: heuristics for risky ClusterRole/Role rules (wildcards, escalate/bind/impersonate) and a sample of ClusterRoleBindings.

StorageClasses, PVCs, PVs, and VolumeAttachments with phase flags (Pending/Lost, Released/Unbound). Omit namespace to list PVCs in all namespaces. Read-only.

List NetworkPolicies with pod selector summary, ingress/egress rule counts, and a possible default-deny ingress heuristic. Omit namespace for all namespaces. Read-only.

Read-only summary: Kyverno PolicyReports / ClusterPolicyReports and Gatekeeper constraint violation counts when those API groups exist.

Markdown reference for DCGM and common GPU-related PromQL examples (no live query). Pair with prometheus_query for execution.

Generic resource lister — fetches any Kubernetes resource by kind and API version.

Fetch logs for a specific pod/container.

Return the full spec of a Kubernetes resource as YAML or JSON. Supported kinds: pod, service, deployment, replicaset, statefulset, daemonset, configmap, secret, ingress, namespace, node, persistentvolumeclaim.

Fetch recent Kubernetes events, optionally filtered by namespace, resource, or type.

Fetch CPU/memory usage for pods via the metrics API (requires metrics-server).

Comprehensive pod diagnosis: status, conditions, container states, events, and recent logs in one call.

Generate a Mermaid diagram showing deployments, services, and pods within a namespace and their connections via selectors.

Generate a Mermaid diagram mapping services to their backend pods via label selectors.

Generate a high-level Mermaid diagram of nodes, namespaces, and pod distribution across the cluster.

Full KubeOrch-style architecture diagram showing ALL resource types in a namespace: Ingress → Service → Deployment → Pod, plus ConfigMaps, Secrets, PVCs, and how they wire together via selectors, env refs, and volume mounts.

Render the cluster topology diagram as a PNG image. Shows nodes, namespaces, and pod distribution as a visual image.

Render a namespace diagram as a PNG image. Shows deployments, services, pods, and their connections.

Render the full KubeOrch-style app architecture as a PNG image. Shows Ingress → Service → Deployment → Pod plus ConfigMaps, Secrets, PVCs.

Render the service-to-pod network map as a PNG image. Shows how services route to backend pods via label selectors.

Render a PNG dashboard of GPU extended resources per node (allocatable) vs sum of pod requests. Requires nodes with GPU resources or workloads requesting GPUs.

Render a PNG bar chart from Prometheus DCGM metric DCGM_FI_DEV_GPU_UTIL. Requires KUBECRAFT_PROMETHEUS_URL and DCGM exporter.

PNG sparkline from a Prometheus range query (default DCGM_FI_DEV_GPU_UTIL, last hour). Requires KUBECRAFT_PROMETHEUS_URL.

List all kubeconfig contexts with cluster, user, namespace, and which is active.

Switch the active kubeconfig context. All subsequent tool calls will use this cluster.

List Helm releases with status, chart, app version, and revision number.

Show user-supplied values for a specific Helm release.

Return KubeCraft MCP server stats: total calls, errors, average latency per tool, and uptime.

Detect whether the cluster is OpenShift and return OCP version, console URL, and registered OpenShift API groups. Works on any cluster.

List OpenShift Routes with host, path, TLS termination, and target service. Returns a compatibility notice on non-OpenShift clusters.

List OpenShift Projects with status, requester, and display name. Returns a compatibility notice on non-OpenShift clusters.

List OpenShift BuildConfigs with build strategy, source type, git URI, and build count. Returns a compatibility notice on non-OpenShift clusters.

Detect whether ArgoCD is installed and report application/project counts and REST API config status.

List ArgoCD Applications with sync status, health, source repo, revision, and target namespace.

List ArgoCD AppProjects with allowed source repos, destination namespaces, and role count.

Detect whether Flux CD is installed and report source/kustomization/helmrelease counts.

List Flux GitRepositories and HelmRepositories with ready status, URL, and artifact revision.

List Flux Kustomizations with ready status, source reference, path, and last applied revision.

List Flux HelmReleases with ready status, chart name, version, release name, and values summary.

Generate a Mermaid diagram of Flux sources flowing into Kustomizations and HelmReleases with ready status colors.

Detect whether Tekton is installed and report pipeline/task/run counts.

List Tekton Pipelines with task count and last PipelineRun status.

List recent Tekton PipelineRuns with status, duration, start time, and task run count.

List Tekton Tasks with step count and parameter count.

List pods that request or limit GPU extended resources (e.g. nvidia.com/gpu) with node, phase, requests, and limits.

Heuristic scan for GPU device plugins, GPU operator workloads, and GPU-related CRDs.

Show current KubeCraft license status, type, expiry, hardware ID, email, and purchase/renewal instructions.

Register or update your email address and optionally apply a new license key. Binds the license to the current machine's hardware fingerprint. Works even when the license is expired.

List DaemonSets with desired/ready/available/updated counts, node selector, and update strategy. Omit namespace for all namespaces. Read-only.

List StatefulSets with ready/updated/current counts, update strategy, and headless service name. Omit namespace for all namespaces. Read-only.

List CronJobs with schedule, suspend status, active job count, and last schedule time. Omit namespace for all namespaces. Read-only.

List Jobs with completion status, duration, and succeeded/failed counts. Omit namespace for all namespaces. Read-only.

List installed CSI drivers with their capabilities (attach, capacity tracking, pod info, fsGroup policy) and per-node registration status from CSINode objects.

List IngressClass resources with controller, default status, and parameter references. Identifies which IngressClass is the cluster default.

Detect installed service meshes (Istio, Linkerd, Cilium, Contour, Kuma, Consul Connect) by API groups. For Istio/Linkerd: shows namespace injection labels and sidecar container counts from a pod sample.

Detect the cluster CNI plugin from node annotations, known DaemonSets (Calico, Cilium, Flannel, AWS VPC CNI, Azure CNI, Weave, Antrea, etc.), and pod CIDRs.

Check CoreDNS deployment health: replica readiness, Corefile configuration (plugins, forwarding), kube-dns Service ClusterIP and endpoint count. Flags single-replica or degraded states.

Resolve a hostname from inside the cluster by execing nslookup/getent/python in a running pod. Compares cluster-DNS view to external DNS to detect split-horizon issues.

Inventory every CRD with group, kind, scope, served versions, stored version, age, and an owner hint (Helm release / OLM CSV / known operator). Optional count_instances includes per-CRD instance counts (slower).

Detail on a single CRD: full version list, conversion strategy, printer columns, subresources, validation summary, instance count per version, and risk flags (no schema, preserveUnknownFields, multi-version with no conversion webhook).

Cluster-wide CR census: count instances of every CRD. Catches operator bloat — thousands of abandoned CronJobs, sprawling VirtualServices, stale ApplicationSets. Returns top N by count.

Detect operators running on the cluster by combining signals: known API groups, OLM ClusterServiceVersions, Deployment naming patterns (*-operator, *-controller, *-manager), and operator-style labels. Reports each operator's source, managed CRDs, ServiceAccount, deployment health.