GKE AI Workload Migration Guide
Guides moving existing AI inference workloads from Cloud Run or Gemini APIs onto self-hosted GKE using gcloud and kubectl.
- Repository
- google/skills
- Stars
- 21,782
- Licence
- Apache-2.0
- Files
- 2 files
What it does
When installed, the agent acts as an interactive migration architect for an AI model server that already runs somewhere else (such as Cloud Run, the Gemini API, or a custom VM) and needs to move to self-hosted inference on Google Kubernetes Engine. It works through discovery, design, implementation and validation phases, reads existing configurations with read-only gcloud commands, writes Kubernetes manifests to disk, and provides gcloud and kubectl commands for the user to run, with a default vLLM serving setup and Gateway API load balancing. It does not handle brand new GKE deployments (it points to a separate gke-inference skill for those) and steps aside if the user wants full automation through the Gemini Cloud Assist MCP server.
Great for
- Engineers moving an existing Cloud Run or Vertex AI model endpoint to self-managed GKE GPU infrastructure
- Teams wanting a step-by-step, reviewable migration with manifests written before any changes are applied
- Developers who prefer native gcloud and kubectl over automated infrastructure tooling
Skip it if
- Deploying a new model server on GKE with no existing workload to migrate: the skill itself says to use gke-inference instead
- Users who want an agent to change cloud resources automatically: the skill hands off to the Gemini Cloud Assist MCP server for that
- Non-technical users without a Google Cloud project, GKE experience or GPU quota, since cluster and billing changes carry real cost and outage risk
Try asking
- “I have a model running on Cloud Run with a Gemini-based endpoint. Help me plan a migration to GKE inference.”
- “Size the GPU node pool I need to serve my existing vLLM-compatible model on GKE.”
- “Generate the Kubernetes manifests and kubectl steps to cut traffic over from my Cloud Run service to GKE.”
- “Help me validate my GKE inference deployment before switching production traffic.”
Install in 3 steps
- 1
Download the skill
The ZIP contains the 2 original files, the Apache-2.0 licence and an attribution note.
google-cloud-solution-guided-gke-ai-migration.zip - 2
Add it to your agent
Claude Code (all projects)
mkdir -p ~/.claude/skills && unzip google-cloud-solution-guided-gke-ai-migration.zip -d ~/.claude/skills/Claude Code / Agent SDK (this project only)
mkdir -p .claude/skills && unzip google-cloud-solution-guided-gke-ai-migration.zip -d .claude/skills/Claude apps (web/desktop): Settings → Capabilities → Skills → Upload skill, and choose the ZIP.
- 3
Ask for the task
No special command needed — the agent reads the skill's description and loads it when your request matches. Try one of the prompts above.
Get better results
- Give the agent read-only permission first so it can inspect your existing Cloud Run or Vertex configuration before proposing anything.
- Review every generated manifest and command yourself; the skill is designed for you to run mutating commands unless you explicitly ask the agent to execute them.
- Mention whether you want to use the Cloud Assist MCP server or stay manual, since the skill routes requests differently depending on that choice.
What's inside
- references/attribution.md
- SKILL.md
Redistributed unmodified from google/skills under the Apache-2.0 licence. The assessment was drafted with AI from the skill's files on 2026-10-29. Skills can instruct your agent to run code — read SKILL.md and any scripts before installing.