Clojurians Log v2

Just close and move on. Then consider if you care about the original issue enough to prompt a chatbot yourself.

In OSS, you never owed nothing to nobody, but in 2026 it’s quadruply so.

yeah I guess I should let go of the idea of having a tidy PR/issue status and accept things just heaping up hopelessly now

Good question

Yeah I'd close and/or fix/revisit yourself.

Even the act of doing that will make you meditate on the changes a little more. I don't think people who slop PRs should get any attribution. After all it's not their work.

Also it might discourage future slop drive bys

Being curious, I’d love to know the PR creator’s motivation.

feel free to engage and ask them. https://github.com/babashka/process/pulls the README PR reads too much as slop to me

Perhaps some CV-motivated boosting of github metrics?

GitHub user description: Just a guy looking to help find bugs and improve projects :)

So is a criteria that the author using AI fully completely and wholly understands their PR? Would that be a good checkbox to add to your PR template? And/or maybe some sort of motivation checkbox? Not sure what it would say, but could spitball some ideas if this is a path of interest.

For prose, I usually find AI-generated prose annoying. I personally don’t mind human written AI-verified prose though.

You all are applying too much effort to justify stopping somebody who applies no effort at all. We hated and blocked bots for years. What has changed now? Is it just because there is a person sitting behind a bot now and relaying its actions?

A valid point!

The PRs are rather small. Seems that at least he is using better prompting?

Yeah, if the author hadn’t used the word Codex in the commit description, I’m not sure I could have detected AI usage.

I have seen this in other projects. People seem to do these “drive-by” PRs. I am not even sure this guy knows what babashka is, let alone about the issue he is fixing. As soon as you see “136 PRs in 85 repositories” on the profile, ignore it. These people don’t honour your time.

Some people seem to think a PR is the new issue and it’s horrible for certain open-source projects as they get swamped with slop PRs.

SlateDB recently went back to a closed contributors model because of that reason. You first need to gain the trust of the community before you are allowed to open a PR.

My first thought was that it's possibly a hacker adjacent thing, that could be trying to push malicious code.

@didibus not a bad hypothesis. 1. Spin up accounts and gain reputation and trust by spamming legit PRs. 2. Make increasingly larger PRs. 3. Sneak in malicious PRs. 4. ???? 5. Profit.

Seems like an old account. A lot of people have no visible presence on GitHub around the tech they use at work.

Would it be bad to invite him into the Clojure forum so that he can talk to people?

Then I’d also invite this guy here, he seems really interested in all of my Clojure blog posts 🙃

> Seems like an old account. I wouldn’t read too much into that accounts get sold for star framing etc.

If I learned anything over the years, is that “hammock time” is super important. What’s great about it, you don’t have to explain to anyone what exactly happens in the hammock. i.e., the first instinct to a problem should be to ignore it. Or at least pretend that you’re dealing with, but in the hammock. In my experience, if you wait long enough, the problem soon resolves itself - in most cases people find that it actually never was a problem. Sometimes, it is still a problem, yet it becomes someone else’s headache. “If you wait by the river long enough, the bodies of unresolved GitHub issues will float by”

I'm thinking, like others, that this is a good preamble to a future supply chain attack

Protect your energy and time. Close and move on.

If you’re semi-open to getting PRs from AI, you could add an agent instruction. Mine warns the user that it’s likely to get rejected for UI/UX, but with the kind of architectural decisions I’m making I’ll probably broaden it a bit if I get more people trying to contribute https://github.com/Naomarik/sova/blob/master/agents/contributing.md

Maybe a good screening would be: post your issue on Github, then jump onto Clojurians Slack to explain who you are, have a chat about the issue and then we can discuss a PR (or not)

and maintain a list of trusted contributors (like some other projects like ghostty do)

Or maybe just Github Discussions instead of Clojurians Slack

but that would be too easy for fully automated bots

Yeah what you said sounds sane. Someone would have to build up a lot of trust for me to even begin assessing PRs to my project. I have no idea how the authors of other harnesses can have >100 contributors going on.

this is how much consideration I’ve given to a relatively simple to understand bug fix

With all the talks about “cognitive surrender” it’s easy for someone to just lob a +8000 PR diff with a few mins of effort. And I’m assuming that’s what everyone is doing until I get trust.

I posted this: Hi @Sinkleberg, I noticed you submitted several PRs to several of my projects in a short amount of time. These projects seems to be unrelated which raises the question what is your motivation in submitting these PRs. Have you used Clojure before? Are you really using these libraries and in what capacity? As it’s harder and harder to build trust in this age of LLM-drive by PRs, I encourage you to hop on to Clojurians Slack and introduce yourself to me, e.g. in the borkdude-misc channel.

Sounds nice

https://github.com/borkdude/.github/blob/main/CONTRIBUTING.md

I suppose a smart enough ai can just do all above to fit this contributing md guide and pretend human to build fake trust? Focus on the pr content and what it fixes and achieves maybe a better approach regardless of the creator(whether bot or human)’s intension/motive.

we’ll see

I like this contributing guideline. Seems reasonable

I like that it is thoughtful and kind. This matches the spirit of the Clojure community as I understand it.

> Fair question. I look for open-source issues to help with as a hobby, and Codex wrote and tested this change. What brought me to rewrite-edn was https://github.com/borkdude/rewrite-edn/issues/37, specifically editing a tagged value without losing the tag or the comments around it. Thanks for the Slack invite.

facepalm

> Codex, answer this comment.

closed his PRs

Well, you certainly tried!

I told you are making it an amplification attack against your time.