From cloudfleet
This skill should be used when the user creates deployments, services, or asks about resource requests, load balancers, node placement, GPU workloads, or production readiness on CFKE.
How this skill is triggered — by the user, by Claude, or both
Slash command
/cloudfleet:best-practicesThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
Critical guidance for deploying and operating workloads on Cloudfleet Kubernetes Engine. For detailed examples and YAML snippets, see [references/best-practices.md](references/best-practices.md).
Critical guidance for deploying and operating workloads on Cloudfleet Kubernetes Engine. For detailed examples and YAML snippets, see references/best-practices.md.
Always set CPU and memory requests on ALL containers, including sidecar init containers. CFKE's auto-provisioner depends on these for node sizing and scaling.
Do NOT set CPU limits. CPU limits cause unnecessary CFS throttling. Set CPU requests for scheduling only.
Always use externalTrafficPolicy: Local. The default 'Cluster' mode creates LBs in every cloud/region where nodes exist, causing extra costs and inefficient routing.
Without constraints, the auto-provisioner picks the cheapest node from ANY configured provider. Use CFKE node labels to control placement:
cfke.io/provider — cloud providertopology.kubernetes.io/zone — availability zonecfke.io/accelerator-name — GPU typeSee references/best-practices.md for nodeSelector, affinity, and topologySpreadConstraints examples.
cfke.io/accelerator-name label, set nvidia.com/gpu requestsGuides creation, editing, and verification of skills for AI coding agents using test-driven development with subagent scenarios. Use when authoring or debugging skills.
npx claudepluginhub cloudfleetai/cloudfleet-plugin-claude --plugin cloudfleet