Public repositories seem to be getting more AI-generated pull requests, some of which are low-effort and haven’t been manually reviewed.

We built Haystack Slop Detector to scan PRs and flag changes that look like they were made without reviewing them first. We support AI use, but we think it’s important to surface a lack of polish before it becomes a maintainer’s problem.

We'd love feedback, especially from maintainers and people building agentic workflows!