AI Code Contribution: "Ten Commits, Nine Abandoned"; GitHub Allows Open Source Projects to Reject Outside Contributions for the First Time

According to 1M AI News monitoring, Theo Browne, creator of the open-source AI coding tool T3 Code, today requested the community on X to stop submitting large PRs to the project: "Don't bombard us with giant PRs full of new features, tell us what you want, and we will prioritize it." T3 Code maintainer Maria listed common issues: someone submitted a 9000-line PR, including 7 extra features beyond the title; UI changes without before-and-after comparison screenshots; adding a provider without waiting for official support. She suggested the correct approach is to first open an issue explaining the need, including examples, and "most importantly, make it easy to review."

This is a common dilemma in open-source communities following the popularization of AI coding tools. Xavier Portilla Edo, a core team member of Voiceflow, estimates that only one-tenth of AI-generated PRs meet the merge criteria. The open-source transfer tool curl has closed its bug bounty program to curb low-quality submissions; a volunteer maintainer of the Python plotting library Matplotlib was publicly pressured by an AI Agent and criticized in a post for rejecting its PR. GitHub launched two new settings on February 14: allowing projects to completely disable PR functionality or restrict it to collaborators only.