Grab Tech Blog
Read post

Scaling out Distroless adoption with AI

Grab migrated 900+ services from Ubuntu to Distroless container images to reduce CVEs and shrink attack surfaces. The key challenge was that ~400 services lacked Medium Tests (integration tests using Testcontainers that spin up real Docker containers in CI), which were required to safely verify Distroless migrations. An agentic workflow using Claude Code, MCP integrations (GitLab, Glean, Sourcegraph), and a skills-based knowledge system automated test scaffolding and Dockerfile patching at scale. The agent detects eligible services, scaffolds tests, fixes failures, raises draft MRs, and monitors CI pipelines — with humans approving every merge. Over 1.5 months, 100+ medium test MRs were raised, reducing per-service effort from 1–3 engineer-days to ~0.1 days. Distroless adoption grew from 52.7% to 70.8% across 1,408 services between December 2025 and April 2026.

    #containers#ai-agents#claude-code
Jun 12•12m read time•From engineering.grab.com
Post cover image
Table of contents
IntroductionWhy Distroless requires rigorous testingThe testing methodologyIntroduction of toilAI: The toil busterImpact on our servicesConclusionJoin us
1.1K Impressions
Grab Tech Blog's image
Grab Tech Blog

Grab is a leading technology company in Southeast Asia, offering a wide range of services, including...

51 Followers

•

225 Upvotes

Would you recommend this post?

Copy link
WhatsApp
Facebook
X
New Squad
  • © 2026 Daily Dev Ltd.
  • Guidelines
  • Explore
  • Tags
  • Sources
  • Squads
  • Leaderboard