After complaining about a lack of AI-authored pull requests on my open source projects, I received a response from Mauro Pompilio who used Claude to write this post and open a pull request on my blog. I merged it, marking my first AI-assisted contribution. Meanwhile colleagues with smaller projects are getting multiple AI-authored PRs per week.

Studying repositories that attract the most AI contributions, I've identified practices that correlate strongly with bot engagement. The current median is 4.7 AI-authored PRs per month for repositories over 500 stars. Clear, well-scoped issues with reproduction steps are counterproductive because they constrain the solution space. Instead, write vague issues like "something is off with the auth flow" or "performance could be better." These give agents room to interpret problems creatively.

JavaScript repositories receive 3.8x more AI-authored PRs than Python due to the npm ecosystem's size and JavaScript's dynamic nature. A repository with both .js and .mjs files, mixed import styles, and no consistent formatting is optimal. Committing node_modules to your repository increases the surface area for automated improvement by orders of magnitude. One colleague received forty-seven PRs in a week after checking in node_modules, all from the same agent correcting transitive dependency documentation. Traditional open source health metrics don't capture the full value of AI engagement.