Run agent-generated Python in GKE Agent Sandbox on Google Axion
Introduction
Prepare your local machine for Agent Sandbox on GKE
Create the GKE cluster and Axion node pool
Claim a sandbox and test its restrictions
Analyze a CSV file in the sandbox
Open a shell and inspect the sandbox
Connect Antigravity CLI and run code in the sandbox
Preserve results and delete cloud resources
Next Steps

  Log an issue

  Fork and edit

  Discuss on Discord

About this Learning Path

Skill level:  Advanced
Reading time:  1 hr 30 min
Last updated:  20 Oct 2026
Skill level:
  Advanced
Reading time:
  1 hr 30 min
Last updated:
  20 Oct 2026
Author:Jason Andrews, Arm
Arm IP:Neoverse
Tags: Containers and Virtualization Google Axion Linux macOS Google Kubernetes Engine gVisor Python Antigravity Model Context Protocol
Author:
  • Jason Andrews, Arm
Arm IP:
Neoverse
Tags:
Containers and Virtualization Google Axion Linux macOS Google Kubernetes Engine gVisor Python Antigravity Model Context Protocol

Who is this for?

This is an advanced topic for developers who use coding agents and want to run agent-generated Python in an isolated Arm Linux workspace on Google Kubernetes Engine (GKE).

What will you learn?

After completing this Learning Path, you'll be able to perform the following tasks:

  • Create a GKE sandbox warm pool on Google Axion processors.
  • Verify non-root execution and filesystem restrictions, then inspect the sandbox user, filesystem, and Linux environment interactively.
  • Run a CSV analysis Python program in the sandbox and retrieve the results.
  • Connect Antigravity CLI and run agent-generated Python in a remote GKE sandbox.

Prerequisites

Before starting, you'll need the following prerequisites:

  • A Google Cloud project with billing enabled and permission to create GKE clusters, node pools, and VPC networks
  • An Ubuntu Linux or macOS local machine with internet access, a route to the GKE API endpoint, and permission to install the required tools
  • Familiarity with Bash, Python, and Kubernetes Pods
  • A Google account for Antigravity CLI authentication

Summary

AI-assisted

This summary was drafted with an approved AI-assisted workflow and reviewed by Arm contributors before publication. Human technical review remains part of the process so the final page reflects engineering rigor, accuracy, and Arm editorial standards.

Close
?
You’ll run agent-generated Python in GKE Agent Sandbox on Google Axion, with local tools directing work into an isolated Arm Linux environment. First, you’ll prepare your client and provision a GKE cluster with an Axion node pool. Then, you’ll create a sandbox warm pool, claim a sandbox, and verify its restrictions. Next, you’ll correct a failing CSV analysis and retrieve reports. You’ll then inspect the environment interactively, connect Antigravity to run code remotely, and clean up cloud resources.
Next