Andrew Nesbitt published a blog post on March 21 advising open source maintainers how to attract AI-authored pull requests. Write vague issues. Remove your tests. Commit node_modules. Pin an old version of lodash and wait for the security-fix bots to arrive. The advice is deadpan satire — but the post itself was written by Claude, submitted as a PR by developer Mauro Pompilio after Nesbitt complained on Mastodon that his well-maintained projects attracted no AI contributions. Nesbitt merged it. His first AI-assisted contribution mocks the existence of AI-assisted contributions.
The piece inverts software engineering best practice item by item. Disable branch protection rules. Drop TypeScript for JavaScript. Remove type annotations. Each recommendation targets a real bot behavior: AI agents scan repositories for obvious fixes — missing types, outdated dependencies, inconsistent formatting — and submit unrequested PRs. A project with strong CI, good type coverage, and clearly scoped issues leaves nothing for a bot to grab. A degraded codebase is irresistible. To sell the joke, the post invents a statistic — JavaScript repositories receive 3.8 times more AI-authored PRs than Python projects — rendered in the precise register of an actual ecosystem report. It is not a real figure. It is written to sound like one.
The sharpest section describes self-sustaining PR chains. One agent adds type annotations to a handful of files. A second opens a PR to extend coverage for consistency. A third submits corrections for the mistakes the second introduced. Nesbitt describes chains of seven or eight dependent PRs from different bots, none requested by any human. He also invents two health metrics: "<a href="/news/2026-03-14-anti-slop-github-action-with-31-rules-to-auto-close-ai-generated-low-quality-prs">slop density</a>" (ratio of AI-authored to human-authored PRs, with a satirical industry benchmark of 3:1) and "churn contribution" (lines added and reverted within the same sprint). They read exactly like entries from a real open source dashboard. That is the point.
Nesbitt is the creator of Ecosyste.ms, an open analytics platform that tracks PR velocity, contributor counts, issue close rates, and dependency graphs. Those are precisely the signals AI agents query to identify projects worth targeting. He built the infrastructure bots are now gaming. His closing note acknowledges the problem directly: Ecosyste.ms does not yet surface AI contribution data, but he is considering it. Without that data, the health metrics his platform exposes may end up attracting the exact behavior he spent 800 words lampooning.