---
title: "Comment by Kartik N V J K on \"Code Review Starts with What Must Stay True\" | Hashnode"
description: "django-oscar #4594 reading like a pure lock-ordering fix is exactly why it tests the method well. I keep coming back to the ArchMap step, though. If the model r"
url: "https://hashnode.com/posts/code-review-starts-with-what-must-stay-true/6ade12731c09b9fd0c9f51b4/comment/6adf3fbef218f0c2c97015d7"
---

[Discussion](https://hashnode.com/posts/code-review-starts-with-what-must-stay-true/6ade12731c09b9fd0c9f51b4)

Comment [Code Review Starts with What Must Stay True](https://hashnode.com/posts/code-review-starts-with-what-must-stay-true/6ade12731c09b9fd0c9f51b4)

[Kartik N V J K](https://hashnode.com/@kartik-nvj)

[Oct 26](https://hashnode.com/posts/code-review-starts-with-what-must-stay-true/6ade12731c09b9fd0c9f51b4/comment/6adf3fbef218f0c2c97015d7)

django-oscar #4594 reading like a pure lock-ordering fix is exactly why it tests the method well. I keep coming back to the ArchMap step, though. If the model records basket order for the sorted loop, L2 passes for the wrong reason. Who checks the ArchMap against the code before the Laws run, the reviewer or another model?

[Hiroyuki Nakahata](https://hashnode.com/@iroha1203)

[Oct 26](https://hashnode.com/posts/code-review-starts-with-what-must-stay-true/6ade12731c09b9fd0c9f51b4/comment/6adf3fbef218f0c2c97015d7#reply-6adf51c82498241548ffab31)

Thanks for the comment! That’s exactly the issue. I’m currently developing a SKILL to help agents produce more accurate ArchMaps from code. The idea is to improve accuracy by having multiple agents observe the code using that SKILL.

Problems that those observations cannot capture, such as runtime race conditions, are outside the scope of what AAT can establish from the resulting ArchMap.