Kosei Horikawa, Hao Li, Yutaro Kashiwa, Youssef Esseddiq Ouatiti, Mohammed Sayagh, Bram Adams, Hajimu Iida, Ahmed E. Hassan
AI coding agents such as Claude Code, GitHub Copilot coding agent, and Cursor now commit directly to open-source repositories. Empirical studies of these agents identify their commits through authorship signatures, such as bot accounts and Co-authored-by trailers. These signatures are optional, can be removed, and are often attached to pull requests rather than to individual commits. Hence, it is unclear whether Agent commits can be told apart from Human commits once the signatures are gone. We investigate the signals and limits of signature-free detection using 7,837,946 commits from 1,226 repositories, covering five coding agents and Human commits. After signatures are masked, detectors rank Agent commits above Human commits with a ROC AUC of up to 0.859, and most of this signal comes from message verbosity and change size rather than from the code itself. On the shared sample, gpt-6-sol without fine-tuning reaches AUCs similar to XGBoost and TF-IDF. At the observed Agent share, however, the best detector on the full test set reaches a precision of only 0.332. Detection transfers poorly to held-out Claude Code and Copilot, with TF-IDF reaching an AUC of only 0.524 for Claude Code. Temporal changes also depend on the agent, e.g., Copilot AUC drops and later recovers, while more training history helps Claude Code but not Copilot. These results indicate that signature-free detection can prioritize commits for review but cannot replace recorded attribution, and that studies relying on signature-based labels should treat them as a lower bound on agent activity.
Kosei Horikawa, Hao Li, Yutaro Kashiwa, Bram Adams, Hajimu Iida, Ahmed E. Hassan
Agentic coding tools, such as OpenAI Codex, Claude Code, and Cursor, are changing how software is developed. These AI-powered systems function as autonomous teammates capable of planning and executing complex development tasks. Agents have become active participants in refactoring, a key practice for improving internal code quality without altering observable behavior. Despite their increasing adoption, there is little empirical understanding of how agentic refactoring is used in practice, how it compares to human-driven refactoring, and what impact it has on code quality. To address this gap, we present a large-scale study of AI agent-generated refactorings in real-world open-source Java projects, analyzing 51,412 refactoring instances across 12,256 pull requests and 14,998 commits based on the AIDev dataset. We find that refactoring is a common and intentional activity, with agents explicitly targeting refactoring in 26.1% of commits. The subsequent analyses of refactoring types, purposes, and code quality focus on the 3,907 commits (32,611 instances) with explicit refactoring intent. High-level refactorings form the largest category overall; however, agentic refactoring contains a slightly larger share of low-level, consistency-oriented edits than human refactoring (27.6% vs. 24.4%), and its two most frequent operations are Rename Variable (7.6%) and Change Variable Type (6.8%). Additionally, the stated motivations behind agentic refactoring focus overwhelmingly on internal quality concerns (52.5% maintainability and 28.1% readability), although the measured metric improvements from these frequent low-level refactorings remain limited. Finally, code quality metrics show that agentic refactoring yields small but statistically significant improvements in structural metrics, particularly for medium-level changes, reducing class size and complexity.
Continuous Integration and Delivery (CI/CD) are fundamental to modern software development, yet configuring pipelines such as GitHub Actions (GHA) workflows remains complex and error-prone. Recent work has explored Transformer-based models for GHA workflow recommendation, but existing approaches typically model workflow steps as a single homogeneous generation task. This formulation overlooks a key distinction: reusable Actions are structured, ecosystem-level components, whereas project-specific Scripts encode local build, test, and deployment semantics. Consequently, current models struggle to capture the different forms of knowledge required for accurate workflow completion. In this work, we present FLOWSTEP, a framework for GHA workflow step completion that explicitly distinguishes between Action and Script generation. FLOWSTEP combines large language models (LLMs) with contrastive learning to better capture semantic relationships among workflow contexts, Actions, and Scripts. Evaluated on established benchmarks, FLOWSTEP improves end-to-end workflow step prediction accuracy from 38.23% to 51.09%. Beyond prediction accuracy, we further study whether LLM-as-a-judge can provide a more semantic evaluation of generated workflow steps. By comparing LLM-based assessments with human annotations, we find that the best-performing model configuration reaches substantial agreement with human ground truth (Krippendorff's alpha = 0.701), suggesting that LLM-as-a-judge can serve as a scalable complement to manual evaluation.