Five AI Agents After the Code: What GitLab Duo Did in a Real Pipeline
Quick answer: I put five GitLab Duo agents (plan, review, fix, secure, release) around a small Python service and let them work after the code was written. They planned a feature in about two minutes, sent one fix back for review three times, diagnosed a pipeline race, and cut container scan findings from 166 to 1. They never merged anything: I did, and I made one mistake doing it. Writing the…
Five GitLab Duo agents, specializing in plan, review, fix, secure, and release, were deployed around a Python service after the code was written. Their tasks included planning a feature in under two minutes, fixing issues multiple times, diagnosing pipeline conflicts, and reducing container scan findings from 166 to just 1. The agents never merged any code, but the author did, making one mistake in the process.
The story details how these AI agents can take on various aspects of building and maintaining a small service, with the slowest part being the work that follows the initial coding stage.
Written by urgent.news from Dev.to's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
