Five AI coding agents produced PRs that exhibit distinct descriptive styles.
Differences in PR structure correlate with variations in reviewer engagement, response time, and sentiment.
Merge outcomes vary across agents, linked to how PRs are presented.
The study highlights the influence of PR presentation on human-AI collaboration dynamics.
The research was conducted using the publicly available AIDev dataset and was submitted to arXiv in early 2026.
📖 Full Retelling
The study, titled *How AI Coding Agents Communicate: A Study of Pull Request Description Characteristics and Human Review Responses*, was authored by Kan Watanabe, Rikuto Tsuchida, Takahiro Monno, Bin Huang, Kazuma Yamasaki, Youmei Fan, Kazumasa Shimari, and Kenichi Matsumoto. It was published on arXiv (ID 2602.17084) on 19 February 2026, and investigates how five AI coding agents differ in the way they compose pull request (PR) descriptions on GitHub and how human reviewers respond to those descriptions. The researchers used the AIDev dataset to analyze structural features of PRs, review activity, response timing, sentiment, and merge outcomes, uncovering distinct communication styles and their effects on reviewer engagement and merge rates.
🏷️ Themes
AI-driven software development, Human-AI collaboration, Pull request analysis, Reviewer behavior and engagement, Large language model output evaluation
Entity Intersection Graph
No entity connections available yet for this article.
Deep Analysis
Why It Matters
The study shows that the way AI coding agents write pull request descriptions directly influences how human reviewers engage with them, affecting response time and merge decisions. Understanding these communication patterns is crucial for improving human‑AI collaboration in software development.
Context & Background
AI coding agents use large language models to generate code changes and submit pull requests automatically
Pull request descriptions shape reviewer perception and determine the speed and outcome of code reviews
The study analyzes five different agents to uncover stylistic differences and their effects on review metrics
What Happens Next
Future work will likely focus on developing guidelines for AI-generated pull request wording and integrating automated feedback loops to optimize reviewer interactions. The findings may also inform tool designers to create better interfaces that bridge AI contributions and human oversight.
Frequently Asked Questions
What is a pull request description and why does it matter?
A pull request description summarizes the changes, explains intent, and provides context; it guides reviewers in evaluating the code and can influence how quickly and positively they respond.
How can AI agents improve their pull request descriptions?
By adopting clearer structure, including relevant details, and aligning tone with team conventions, AI agents can reduce reviewer effort, shorten review time, and increase merge rates.
}
Original Source
--> Computer Science > Artificial Intelligence arXiv:2602.17084 [Submitted on 19 Feb 2026] Title: How AI Coding Agents Communicate: A Study of Pull Request Description Characteristics and Human Review Responses Authors: Kan Watanabe , Rikuto Tsuchida , Takahiro Monno , Bin Huang , Kazuma Yamasaki , Youmei Fan , Kazumasa Shimari , Kenichi Matsumoto View a PDF of the paper titled How AI Coding Agents Communicate: A Study of Pull Request Description Characteristics and Human Review Responses, by Kan Watanabe and 6 other authors View PDF HTML Abstract: The rapid adoption of large language models has led to the emergence of AI coding agents that autonomously create pull requests on GitHub. However, how these agents differ in their pull request description characteristics, and how human reviewers respond to them, remains underexplored. In this study, we conduct an empirical analysis of pull requests created by five AI coding agents using the AIDev dataset. We analyze agent differences in pull request description characteristics, including structural features, and examine human reviewer response in terms of review activity, response timing, sentiment, and merge outcomes. We find that AI coding agents exhibit distinct PR description styles, which are associated with differences in reviewer engagement, response time, and merge outcomes. We observe notable variation across agents in both reviewer interaction metrics and merge rates. These findings highlight the role of pull request presentation and reviewer interaction dynamics in human-AI collaborative software development. Subjects: Artificial Intelligence (cs.AI) ; Software Engineering (cs.SE) Cite as: arXiv:2602.17084 [cs.AI] (or arXiv:2602.17084v1 [cs.AI] for this version) https://doi.org/10.48550/arXiv.2602.17084 Focus to learn more arXiv-issued DOI via DataCite (pending registration) Submission history From: Kan Watanabe [ view email ] [v1] Thu, 19 Feb 2026 05:06:31 UTC (98 KB) Full-text links: Access Paper: View a PDF...
