Hi HN, I’ve been working on a solution to something I personally view as a problem: increasing amounts of AI Slop in PRs.

I’ve noticed that some contributors on GitHub have been using AI tools (which are great!) to make low-quality contributions to public repos just for the sake of it. These contributions often have duplicate code, hallucinated functions, and other signature problems introduced by AI tooling. This leaves the burden on the maintainers to slog through long AI-generated PRs.

I think this is wrong. There should be a mechanism to flag low-effort PRs that contain AI slop, so that maintainers can skip over reading them.

I rigged up a simple AI slop detector here: https://haystackeditor.com/slop-detector

For fun, I included some AI slop examples and a “can you guess whether it’s slop” game.

Let me know what you think!