Post

i don't mind people using agents to create PRs, but please at least write the description and replies yourself 🙏

4 reposts81 likes

If I smell AI I just close it. It's a layer of the author's workflow I don't want to see or deal with. Fake it until you make it. I might start missing those good old everything-in-the-title-skip-template contributions.

It's the replies that really get me. If I've asked a question, I want to hear from the human. If I want to hear from an agent I can ask one myself.

I think it's less about whether it is AI generated and more about lack of understanding of the content. Does the person writing that comment know what they are talking about? Do they care whether their comment is coherent and understandable to the person reading it.

Yup exactly this! And often the answer is no, unfortunately. Same benefit of writing things down on paper, it forces you to understand it better

maybe you should start minding? if it's OSS contribution i'm afraid people can always game their stats, and letting an agent do it for them while having the least possible personal involvement is the current easiest way to game it

I’ve been working on a personal project with AI generated PRs, and it’s so bad to read. It’s better to just say “no description” 😵‍💫 It’s impressive how hard it is to read. Even if I say use bullet points, it picks the least important parts of the PR to call out

i often just skip reading the description now and go straight to the code. it does a better job of explaining 😅

Unfortunately you have to understand what's in the PR to write your own descriptions and replies, so you do mind if people use agents to create PRs

It's not about the output method, it's about the output itself, and how it will be consumed. ("AI-generated" and "AI-slop" don't have to be the same thing.) What's appropriate in one repo may not work for another. Contributor's guides are more important than ever.