Type: url
Original: https://arxiv.org/abs/2607.03316
Status: generated
Language: en
Summary
This 2026 paper studies developer responses to CodeRabbit’s agentic code reviews across 31,073 review–feedback pairs from 10,191 pull requests in 239 GitHub repositories, finding 36.4% accepted, 7.3% discussed and 56.3% rejected. Rejections were mainly tied to invalid suggestions (false positives, redundant or out-of-scope comments) and misalignment with developer intent, functional comments were more often invalid than evolvability ones, and lightweight learning-based methods predicted rejection with up to 76% F1.