What problem does it solve? Designing and configuring the Kubernetes networking plane requires deep knowledge of the pod network model, CNI mechanics, Services/kube-proxy, and Calico's architecture, and mistakes like wrong API groups, unsupported eBPF combinations, or HostEndpoint lockouts can break a cluster. ## Core Features & Use Cases - Network Model & CNI Mechanics: Explains the IP-per-pod no-NAT model, the four networking problems, CNI ADD/DEL invocation, conflist chaining, and Services/kube-proxy/CoreDNS behavior. - Calico Architecture & Dataplanes: Covers Felix/BIRD/confd/Typha components, iptables/eBPF/nftables/VPP dataplanes, Tigera operator install, IPAM pools and borrowing, BGP mesh vs route reflectors, and VXLAN/IPIP/native encapsulation decisions. - Policy Data Model: Details NetworkPolicy/GlobalNetworkPolicy on projectcalico.org/v3 with action, order, tiers, HostEndpoint failsafes, and network sets, including OSS vs Enterprise feature gates. - Use Case: When migrating a 200-node on-prem cluster off node-to-node BGP mesh, use this Skill to configure route reflectors, advertise service ClusterIPs with externalTrafficPolicy, and choose native routing over overlay. ## Quick Start Ask the AI to design a Calico IPPool and BGPConfiguration for a cluster that needs routable LoadBalancer service IPs without an external load balancer.