What problem does it solve?
Finding the correct golden base image for building custom GKE node images requires mapping many cluster parameters (GKE version, OS, architecture, accelerators, gVisor, cgroup mode) to Google's published image maps, which is tedious and error-prone when done manually.
Core Features & Use Cases
- Parameter Inference: Translates natural-language context clues (e.g., "I need GPUs" or "hardened nodes") into technical fields like architecture, accelerator flags, and signed-module enforcement, with sensible defaults.
- Image Map Lookup: Fetches the official GKE node-config-to-base-images JSON for the cluster's minor version and filters it by exact version and node configuration.
- Fallback & Validation: Refines searches by toggling cgroup mode or gVisor settings when no match is found, and warns when an unsupported GKE version is provided.
- Use Case: A platform engineer running GKE 1.34.1-gke.2909000 with H100 GPUs asks for the right base image and receives the exact image name, such as gke-1341-gke2909000-cos-125-19216-0-115-c-nvda, ready for custom node image creation.
Quick Start
Ask the agent to find the golden base image for your GKE cluster by describing your GKE version, operating system preference, and any GPU, TPU, or security requirements.