AI coding agents are increasingly acting as autonomous contributors by generating and submitting pull requests (PRs). However, we lack empirical evidence on how these agent-generated PRs differ from human contributions, particularly in how they modify code and describe their changes. Understanding these differences is essential for assessing their reliability and impact on development workflows. Using the MSR 2026 Mining Challenge version of the AIDev dataset, we analyze 24,014 merged Agentic PRs (440,295 commits) and 5,081 merged Human PRs (23,242 commits). We examine additions, deletions, commits, and files touched, and evaluate the consistency between PR descriptions and their diffs using lexical and semantic similarity. Agentic PRs differ substantially from Human PRs in commit count (Cliff’s δ = 0.5429) and show moderate differences in files touched and deleted lines. They also exhibit slightly higher description-to-diff similarity across all measures. These findings provide a large-scale empirical characterization of how AI coding agents contribute to open source development.

Abstract The rapid adoption of large language models (LLMs) like ChatGPT has introduced new dynamics in software development, particularly within pull request workflows. While prior research has examined the quality of AI-generated code, less is known about how developers evaluate, adapt, and integrate these suggestions in real-world collaboration. We analyze 338 pull requests from 255 GitHub repositories containing self-admitted ChatGPT usage, comprising 645 AI-generated snippets and 3,486 developer-authored patches. To support this analysis at scale, we use PatchTrack, an automated classifier that identifies whether AI-generated patches were applied, partially reused, or not integrated. Our findings reveal that full adoption of ChatGPT-generated code is uncommon: the median integration rate is 25%. Qualitative analysis of 89 pull requests with integrated patches reveals recurring patterns of structural integration , selective extraction , and iterative refinement , indicating that developers typically treat AI output as a starting point rather than a final implementation. Even when code is not directly adopted, ChatGPT influences workflows through conceptual guidance , documentation , and debugging strategies . Integration decisions reflect contextual fit , integration effort , maintainer trust , and established pull request review norms rather than serving as direct indicators of code correctness. Overall, this study provides empirical insight into AI-mediated decision-making in collaborative software development, showing that the influence of generative AI extends beyond patch generation to how developers reason about, adapt, and negotiate code during review within pull request workflows. These findings inform the design of AI-assisted tools and support more transparent and effective use of LLMs in practice.

While most forks on platforms like GitHub are short-lived and used for social collaboration, a smaller but impactful subset evolve into long-lived forks, referred to here as variants, that maintain independent development trajectories. Integrating bug-fix patches across such divergent variants poses challenges due to structural drift, including refactorings that rename, relocate, or reorganize code elements and obscure semantic correspondence. This paper presents an empirical study of patch integration failures in 14 divergent pair of variants and introduces RePatch, a refactoring-aware integration system for Java repositories. RePatch extends the RefMerge framework, originally designed for symmetric merges, by supporting asymmetric patch transfer. RePatch inverts refactorings in both the source and target to realign the patch context, applies the patch, and replays the transformations to preserve the intent of the variant. In our evaluation of 478 bug-fix pull requests, Git cherry-pick fails in 64.4% of cases due to structural misalignments, while RePatch successfully integrates 52.8% of the previously failing patches. These results highlight the limitations of syntax-based tools and the need for semantic reasoning in variant-aware patch propagation.

In recent years, the integration of AI tools such as ChatGPT into software development has grown significantly, reflecting broader trends in AI-assisted workflows [8]. These tools have great potential to improve decision making related to software patches in pull requests (PR), which are vital components of collaborative software development. Specifically, developers are using features such as link sharing in ChatGPT to enhance collaborative practices, streamline code reviews, and make more informed patch integration decisions.