Back to Skills

gke-storage

Manages GKE storage, including PVCs, PersistentVolumes, Filestore, and GCS FUSE. Use when configuring GKE storage, creating PVCs, or setting up GCS FUSE on GKE. Don't use for database administration or replication strategies outside volume provisioning context.

15,350stars1,203forksUpdated 7/31/2026

Security Assessment

Safe(95/100)
Security Score95/100

About gke-storage

gke-storage is a reference skill for configuring persistent and file storage on Google Kubernetes Engine (GKE) clusters. It addresses the common challenge of choosing the correct storage primitive and access mode across GKE's storage options by documenting golden-path CSI driver defaults, built-in StorageClasses, and ready-to-adapt manifest examples for PersistentVolumeClaims and volume mounts.

The skill covers block storage via Compute Engine Persistent Disk (ReadWriteOnce), shared NFS via Filestore (ReadWriteMany), mounting Google Cloud Storage buckets as volumes through the GCS FUSE CSI driver, and high-performance Parallelstore. It explains the default StorageClasses (standard-rwo, premium-rwo, standard-rwx, premium-rwx), shows how to define a custom regional-PD StorageClass with volume expansion, and gives PVC examples for databases and shared data. It documents volume expansion by patching a PVC (via kubectl or the preferred patch_k8s_resource MCP tool) and includes best practices such as always enabling volume expansion, using regional PDs and WaitForFirstConsumer binding for production, choosing disk types by workload, and backing up PVCs with Backup for GKE. The GCS FUSE example correctly notes that Workload Identity with storage.objectViewer is required for bucket access.

It targets Kubernetes and GKE platform engineers and application developers who need to provision, size, share, or expand storage for their workloads. As a documentation-and-manifest reference (its only file is SKILL.md), it is advisory rather than executing anything itself, and it explicitly defers database administration and replication strategy to other skills.

FAQ

Which storage types does it cover?

Compute Engine Persistent Disk block storage, Filestore shared NFS, Cloud Storage FUSE bucket mounts, and Parallelstore, along with GKE's built-in StorageClasses and custom StorageClass definitions.

How do I resize a volume?

If allowVolumeExpansion is set on the StorageClass, you patch the PVC's requested storage using kubectl patch or the preferred patch_k8s_resource MCP tool, and Kubernetes resizes the filesystem automatically.

What is required to mount a GCS bucket?

The skill notes you must use Workload Identity so the pod's service account holds storage.objectViewer on the bucket, and the pod needs the gke-gcsfuse/volumes annotation plus the gcsfuse CSI driver volume.

What are the production best practices?

Always enable volume expansion, use regional PDs for HA across two zones, use WaitForFirstConsumer binding, pick disk types by workload, use Filestore for shared access, and back up PVCs with Backup for GKE.

What is it not for?

It is not for database administration or replication strategies outside the volume-provisioning context.

Install gke-storage

Download and extract the skill files to your .claude/skills/ directory.

Quick Setup:

  1. Copy the skill folder to .claude/skills/
  2. Claude will automatically detect and use the skill

Repository

google/skills