Stop bloating open source repos with AI-generated junk just to

The sheer volume of low-effort pull requests hitting major open source projects lately is becoming impossible to ignore. I’ve noticed a massive uptick in "contributions" that are nothing more than LLM-generated boilerplate, superficial documentation fixes, or completely irrelevant code snippets that don't actually solve any existing issues. It’s clear what the motivation is: people are trying to farm GitHub green squares and pad their resumes with a high volume of commits to look "active" to recruiters.

This trend is creating a massive maintenance tax for real maintainers. When you submit a PR that was clearly spit out by a prompt without any human verification, you aren't contributing; you're adding to the noise. A maintainer now has to spend precious time reviewing, debugging, and eventually rejecting code that shouldn't have been sent in the first place. This isn't just a minor annoyance—it’s a direct drain on the energy required to keep critical infrastructure running.

If you are using an LLM to assist your coding, that is perfectly fine. In fact, leveraging AI as a tool for a deep dive into a codebase or to help draft complex logic is a legitimate part of a modern AI workflow. However, there is a massive distinction between using Claude or GPT to assist your thinking and using them to automate the act of "contributing" without understanding the underlying logic.

Here is how you can actually use AI to contribute without being part of the "slop" problem:

1. Verify every single line: If you use an LLM to generate a fix, you must be able to explain exactly why that code works. If you can't walk a maintainer through the logic during a review, you shouldn't have submitted it.
2. Focus on substance over volume: One meaningful, well-tested bug fix is worth more than fifty "typo fixes" or documentation updates generated by a script.
3. Test locally first: AI-generated code often looks syntactically correct but fails in real-world edge cases. Never submit a PR unless you have personally run the test suite and confirmed the changes don't break existing functionality.
4. Avoid "Refactoring" for the sake of it: A common pattern with AI slop is the "AI-style refactor," where the LLM rewrites a perfectly functional block of code into something slightly more "elegant" but actually less readable or performant for that specific project. Unless there is a documented technical debt issue, leave the existing logic alone.

The goal of open source is collaboration and improvement. When we treat it like a farm for resume padding, we degrade the quality of the ecosystem for everyone. If you want to show recruiters you know how to work with AI, show them that you can use it to solve hard problems and navigate complex architectures, not that you can successfully copy-paste from a chat window into a terminal.