Kubernetes Skills (3)

**Quick judgement** Solid, focused skill for managing Kubernetes clusters via Cluster API (CAPI). It offers good coverage for cluster and machine lifecycle (create, inspect, scale) with clear workflows and troubleshooting guidance. Best suited for users who already operate CAPI-based infrastructure and need programmatic cluster lifecycle management. **What it does well** - Manages CAPI clusters end-to-end: listing clusters, machines, machine deployments, machine sets, health checks, and cluster classes. - Creates clusters and MachineDeployments declaratively via `kubectl_apply` with concrete manifest examples. - Scales worker nodes safely through MachineDeployment replicas, with recommended monitoring steps. - Exposes kubeconfig for workload clusters, enabling further operations from other skills or tools. - Provides useful operational workflows (provision, scale, upgrade, troubleshoot) rather than just isolated calls. **Key risks / limitations** - **Infrastructure impact & cost risk**: Creating clusters and scaling MachineDeployments (e.g., AWS-backed) can incur cloud costs and resource consumption if misused or left running. - **Configuration risk**: Misconfigured manifests applied via `kubectl_apply` can lead to failed clusters, broken rollouts, or inconsistent states, especially in production. - **No explicit teardown tooling**: The page doesn’t expose dedicated delete/destroy tools for clusters; cleanup may require other skills or manual actions, increasing risk of orphaned infrastructure. - **Environment assumptions**: Requires a working CAPI setup and kubectl-mcp-server integration; without these, many tools will fail. **Recommended scenarios** - Teams operating **Cluster API-based Kubernetes fleets** who want automated provisioning and scaling of workload clusters. - Platform engineers managing **multi-cluster infrastructure** (especially alongside related skills like `k8s-multicluster` and `k8s-operations`). - SREs and operators needing **introspection and troubleshooting** for CAPI clusters and machines (phases, conditions, health checks, events). - Controlled environments (staging, infra-management namespaces) where cluster lifecycle changes are expected and cost/impact is monitored. Use with extra caution in production and cost-sensitive environments, especially when applying new manifests or scaling up nodes; ensure guardrails (reviewed manifests, quotas, and access controls) are in place.