GKE Storage Manager
Configure persistent disks, Filestore, and GCS FUSE storage for GKE workloads.
The skill discloses the Workload Identity and storage.objectViewer requirement and provides a read-only GCS FUSE example. It does not require confirmation before cluster, StorageClass, PVC, or resize changes, and lacks least-privilege guidance, sensitive-data handling, data-flow transparency, rollback, and recovery procedures, so points are deducted.
The file covers common StorageClass, PVC, GCS FUSE, and expansion paths, but its MCP tool list conflicts with the later use of patch_k8s_resource, the tables are malformed, and there are no preflight checks, abnormal-input handling, tests, or diagnosable failure feedback. Static calibration limits reliability to 10 or below.
Invocation scenarios, storage types, and the boundary excluding database administration or unrelated replication strategies are reasonably clear. However, input prerequisites, version-specific differences, additional non-fit boundaries, Chinese-language support, and mainland-China network reachability are not addressed, so points are deducted.
The skill has useful metadata, headings, sections, examples, and best practices. The repository supplies installation guidance, an Apache-2.0 license, support channels, and an active-development notice. It lacks skill versioning, changelog, explicit maintenance ownership, dependency/install notes, FAQs, and limitation disclosure, and the Markdown tables are structurally broken, so points are deducted.
The YAML and kubectl examples are usable starting points covering block storage, shared file storage, GCS mounts, and expansion. However, there is no execution evidence, complete deployment workflow, environment prerequisite checklist, troubleshooting, or security/cost tradeoff guidance; manual validation remains necessary, so static calibration limits effectiveness to 7 or below.
Evidence consists mainly of the skill's static text and repository README/LICENSE. There are no committed tests, CI coverage, third-party execution records, or specific source citations, so verifiability is limited and remains at or below the static cap of 5.
- The examples can change clusters and persistent data; confirm the target cluster, region, access modes, permission scope, and backup/rollback plan first.
- Verify current GKE, CSI driver, Filestore, and GCS FUSE documentation because this file provides no version pinning or validation tests.
- MCP tool names are inconsistent within the file, so tool availability and parameter correctness cannot be assumed from this document alone.
What does this skill do, and when should you use it?
This skill manages storage configuration for GKE clusters, including PersistentVolumeClaims, PersistentVolumes, StorageClasses, Filestore, and GCS FUSE. It documents Golden Path Autopilot storage defaults and provides examples for custom StorageClasses and common volume types. It also covers volume expansion, regional persistent disks, shared file access, and workload identity requirements for bucket mounts. It is suited to GKE storage provisioning, not database administration or replication strategies outside that context.
Uses Kubernetes manifests and the listed MCP tools to configure or inspect GKE storage resources; explains the use cases for Compute Engine Persistent Disk, Filestore, Cloud Storage FUSE, and Parallelstore CSI; provides YAML examples for custom StorageClasses, block-storage PVCs, shared-file PVCs, and direct GCS bucket mounts; shows how to expand a PVC with kubectl or an MCP resource patch.
- A GKE operator needs a 100Gi ReadWriteOnce PVC for a database and selects premium-rwo with pd-ssd.
- A multi-pod application needs shared read/write files and uses a Filestore-backed ReadWriteMany PVC.
- A machine-learning or logging pipeline needs to mount a GCS bucket directly through the GCS FUSE CSI driver.
- A production workload needs a regional persistent disk replicated across two zones.
- A cluster administrator needs to increase an existing PVC's capacity after enabling volume expansion.
What are this skill's strengths and limitations?
- Covers the main GKE paths for block storage, shared file storage, and object-storage mounts.
- Includes copyable YAML and kubectl examples.
- Documents practical defaults and guidance for expansion, WaitForFirstConsumer, and regional PDs.
- States the Workload Identity and storage.objectViewer requirement for GCS FUSE.
- The source provides no test suite, version compatibility matrix, or troubleshooting workflow.
- The GCS FUSE example requires the user to provide a bucket, service-account permissions, and a suitable GKE environment.
- The scope excludes database administration, a complete backup procedure, and standalone replication strategies.
- The MCP tool names are listed, but their server setup and invocation details are not documented.
How do you install this skill?
Install the repository's skill collection with: npx skills add google/skills. The installer allows selecting specific skills from the repository; the source does not document a separate gke-storage installation command or destination folder.
How do you use this skill?
After installing it in an Agent Skills-compatible client, ask for a concrete storage task such as “Create a 100Gi premium-rwo PVC for a database on GKE” or “Configure a read-only GCS FUSE bucket mount for a GKE Pod.” Execution requires access to a target GKE cluster and the client's Kubernetes MCP tools; the source does not define a client-specific trigger syntax.
How does this skill compare with similar options?
The source does not provide a direct comparison with other storage skills or tools; it only presents kubectl as an option and identifies MCP usage as preferred for PVC patching.