Ask HN: Are PMs and colleagues sending you untested AI PRs that do not work?

Re:Ask HN: Are PMs and colleagues sending you untested AI PR... by pavel_lishin on Tuesday October 07, 2025 @02:11PM (#45506641)

> So now I have to review broken PRs at breakneck speed because everyone has gotten so productive that I've become a bottleneck.

These PRs seem like they would take mere minutes to identify as being slop, no? I guess if you're getting dozens of them, that can be a time-sink.

In any case, it's good that you're a bottleneck; you're more than that, you're a sieve, preventing this crap from getting into the codebase.

I, personally, would be up front: "This was written by AI, with zero tests and zero thought. It doesn't follow best practices, and will not be merged as is. I will be rejecting all PRs of this nature in the future." Then close the PR, and copy-and-paste as needed; after the third one, contact their manager.

[ Parent | on HN ]

Re:Ask HN: Are PMs and colleagues sending you untested AI PR... by JohnFen on Tuesday October 07, 2025 @02:46PM (#45507103)

You don't even need to object to the genAI aspect of it. Submitting untested code as a PR is unacceptable, period, regardless of whether or not a human wrote it.

[ Parent | on HN ]

Re:Ask HN: Are PMs and colleagues sending you untested AI PR... by ailurooo on Tuesday October 07, 2025 @04:51PM (#45508691)

perhaps their pipeline doesn't allow the pm's to easily check their code, or their computers aren't setup for running/compiling code? idk

[ Parent | on HN ]

Re:Ask HN: Are PMs and colleagues sending you untested AI PR... by JohnFen on Tuesday October 07, 2025 @05:04PM (#45508852)

If they lack the tools and/or skills to check their code, then they shouldn't be submitting code.

[ Parent | on HN ]