Scaling out Distroless adoption With AI

What this post added

This post introduces an AI-driven approach to scale Distroless image adoption across Grab's services. It details the technical challenges of runtime failures due to missing dependencies in Distroless images and outlines a 'medium test' methodology for verification. The core innovation is an AI agent that automates the creation of these medium tests for services that don't have them, and then manages a 'Patch-Test-Compare' loop to safely migrate services to Distroless images. Key technical components include the use of Claude Code, skills-based agent design, Model Context Protocol (MCP) for integrations with GitLab and Sourcegraph, and token-efficient communication strategies.