Type: url
Status: generated
Summary
A GitHub blog guide to reviewing agent-generated pull requests, arguing that the ease of approving them is itself the problem and that the part of review which does not get automated is judgement. It sets out five red flags — CI gaming, code reuse blindness, hallucinated correctness, agentic ghosting, and untrusted input in workflows — together with a roughly ten-minute review sequence, four conditions for sending a pull request back to be made smaller, and advice to treat automated review as a prerequisite rather than a replacement. It also gives authors guidance: edit the body, annotate the diff, and review your own agent-generated pull request before tagging others.