CodeRabbit Activities in Open Source Software Projects Dataset

Overview This dataset comprising agent activities collected from engineered open source software repositories. Based on 481 repositories with evidence of CodeRabbit adoption, we extracted 99,454 unique PRs and 2,255 issues. Within these artifacts, we automatically identified instances of CodeRabbit activities, including review of code changes, participation in issue discussions, generation of docstrings, generation of unit tests, resolution of merge conflicts and others. The proposed dataset enables empirical studies on AI-assisted code review, including investigations of agent configuration, participation in review discussions, authored contributions, and the feedback generated during PRs reviews.