How AI Coding Agents Modify Code: A Large-Scale Study of GitHub Pull Requests

• URL: http://arxiv.org/abs/2601.17581v1
• Date: Sat, 24 Jan 2026 20:27:04 GMT
• Title: How AI Coding Agents Modify Code: A Large-Scale Study of GitHub Pull Requests
• Abstract summary: We analyze 24,014 merged Agentic PRs (440,295 commits) and 5,081 merged Human PRs (23,242 commits)<n>Agentic PRs differ substantially from Human PRs in commit count (Cliff's $= 0.5429$) and show moderate differences in files touched and deleted lines.<n>These findings provide a large-scale empirical characterization of how AI coding agents contribute to open source development.
• Score: 0.0
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: 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.

Related papers

• How to Interpret Agent Behavior [56.59836196946289]
  We introduce ACT*ONOMY, a taxonomy for describing and analyzing agent behavior at runtime.<n>By providing a shared vocabulary, ACT*ONOMY helps researchers, agent designers, and end users interpret agent behavior more consistently.
  arXiv  Detail & Related papers  (2026-05-13T14:52:40Z)
• SWE-chat: Coding Agent Interactions From Real Users in the Wild [70.18158706281724]
  SWE-chat is the first large-scale dataset of real coding agent sessions collected from open-source developers in the wild.<n>The dataset currently contains 6,000 sessions, comprising more than 63,000 user prompts and 355,000 agent tool calls.
  arXiv  Detail & Related papers  (2026-04-22T17:08:19Z)
• ClawArena: Benchmarking AI Agents in Evolving Information Environments [61.664633997138004]
  ClawArena is a benchmark for evaluating AI agents in evolving information environments.<n>Each scenario maintains a complete hidden ground truth while exposing the agent only to noisy, partial, and sometimes contradictory traces.<n> evaluation is organized around three coupled challenges: multi-source conflict reasoning, dynamic belief revision, and implicit personalization.
  arXiv  Detail & Related papers  (2026-04-05T17:55:23Z)
• Safer Builders, Risky Maintainers: A Comparative Study of Breaking Changes in Human vs Agentic PRs [0.991629944808926]
  We conduct a comparative analysis of 7,191 agent-generated PRs with 1402 human-authored PRs from Python repositories.<n>We show that AI agents introduce fewer breaking changes overall than humans (3.45% vs. 7.40%) in code generation tasks.<n>Agents exhibit substantially higher risk during maintenance tasks, with changes introducing breaking changes at rates of 6.72% and 9.35%, respectively.
  arXiv  Detail & Related papers  (2026-03-29T05:15:13Z)
• Dynamic analysis enhances issue resolution [53.50448142467294]
  DAIRA (Dynamic Analysis-enhanced Issue Resolution Agent) is an automated repair framework that embeds dynamic analysis into the agent's reasoning cycle.<n> Driven by a Test Tracing-Driven methodology, DAIRA utilizes lightweight monitors to extract critical runtime data.<n>When powered by Gemini 3 Flash Preview, DAIRA establishes a new state-of-the-art (SOTA) performance, achieving a 79.4% resolution rate on the SWE-bench Verified dataset.
  arXiv  Detail & Related papers  (2026-03-23T14:48:54Z)
• How AI Coding Agents Communicate: A Study of Pull Request Description Characteristics and Human Review Responses [6.061536429904841]
  We conduct an empirical analysis of pull requests created by five AI coding agents using the AIDev dataset.<n>We find that AI coding agents exhibit distinct PR description styles, which are associated with differences in reviewer engagement, response time, and merge outcomes.
  arXiv  Detail & Related papers  (2026-02-19T05:06:31Z)
• Why Are AI Agent Involved Pull Requests (Fix-Related) Remain Unmerged? An Empirical Study [5.127121704630949]
  We analyze 8,106 fix related PRs authored by five widely used AI coding agents from the AIDEV POP dataset.<n>Our results indicate that test case failures and prior resolution of the same issues by other PRs are the most common causes of non integration.
  arXiv  Detail & Related papers  (2026-01-29T22:06:58Z)
• Code Change Characteristics and Description Alignment: A Comparative Study of Agentic versus Human Pull Requests [0.0]
  We analyze 33,596 agent-generated PRs and 6,618 human PRs to compare code-change characteristics and message quality.<n>Agents generate stronger commit-level messages but lag humans at PR-level summarization.<n>These findings highlight a gap between agents' micro-level precision and macro-level communication.
  arXiv  Detail & Related papers  (2026-01-24T23:33:07Z)
• Where Do AI Coding Agents Fail? An Empirical Study of Failed Agentic Pull Requests in GitHub [5.808464460707249]
  We conduct a large-scale study of 33k agent-authored PRs made by five coding agents across GitHub.<n>We first quantitatively characterize merged and not-merged PRs along four broad dimensions.<n>Not-merged PRs tend to involve larger code changes, touch more files, and often do not pass the project's CI/CD pipeline validation.
  arXiv  Detail & Related papers  (2026-01-21T17:12:46Z)
• On Autopilot? An Empirical Study of Human-AI Teaming and Review Practices in Open Source [11.412808537439973]
  We investigated project-level guidelines and developers' interactions with AI-assisted pull requests (PRs)<n>We found that over 67.5% of AI-co-authored PRs originate from contributors without prior code ownership.<n>In contrast to human-created PRs where non-owner developers receive the most feedback, AI-co-authored PRs from non-owners receive the least.
  arXiv  Detail & Related papers  (2026-01-20T09:09:53Z)
• Analyzing Message-Code Inconsistency in AI Coding Agent-Authored Pull Requests [5.885226503818935]
  Pull request descriptions generated by AI coding agents are the primary channel for communicating code changes to human reviewers.<n>We analyzed 23,247 agentic PRs across five agents using PR message-code inconsistency (PR-MCI)<n>High-MCI PRs had 51.7% lower acceptance rates and took 3.5x longer to merge.
  arXiv  Detail & Related papers  (2026-01-08T12:31:02Z)
• Holistic Agent Leaderboard: The Missing Infrastructure for AI Agent Evaluation [87.47155146067962]
  We provide a standardized evaluation harness that orchestrates parallel evaluations across hundreds of tasks.<n>We conduct three-dimensional analysis spanning models, scaffolds, and benchmarks.<n>Our analysis reveals surprising insights, such as higher reasoning effort reducing accuracy in the majority of runs.
  arXiv  Detail & Related papers  (2025-10-13T22:22:28Z)
• Impatient Users Confuse AI Agents: High-fidelity Simulations of Human Traits for Testing Agents [58.00130492861884]
  TraitBasis is a lightweight, model-agnostic method for systematically stress testing AI agents.<n>TraitBasis learns directions in activation space corresponding to steerable user traits.<n>We observe on average a 2%-30% performance degradation on $tau$-Trait across frontier models.
  arXiv  Detail & Related papers  (2025-10-06T05:03:57Z)
• Every Step Counts: Decoding Trajectories as Authorship Fingerprints of dLLMs [63.82840470917859]
  We show that the decoding mechanism of dLLMs can be used as a powerful tool for model attribution.<n>We propose a novel information extraction scheme called the Directed Decoding Map (DDM), which captures structural relationships between decoding steps and better reveals model-specific behaviors.
  arXiv  Detail & Related papers  (2025-10-02T06:25:10Z)
• DRBench: A Realistic Benchmark for Enterprise Deep Research [81.49694432639406]
  DRBench is a benchmark for evaluating AI agents on complex, open-ended deep research tasks in enterprise settings.<n>We release 15 deep research tasks across 10 domains, such as Sales, Cybersecurity, and Compliance.
  arXiv  Detail & Related papers  (2025-09-30T18:47:20Z)
• AgentPack: A Dataset of Code Changes, Co-Authored by Agents and Humans [46.56091965723774]
  Fine-tuning large language models for code editing has typically relied on mining commits and pull requests.<n>We present AgentPack, a corpus of 1.3M code edits co-authored by Claude Code, OpenAI Codex, and Cursor Agent.<n>We show that models fine-tuned on AgentPack can outperform models trained on prior human-only commit corpora.
  arXiv  Detail & Related papers  (2025-09-26T05:28:22Z)
• On the Use of Agentic Coding: An Empirical Study of Pull Requests on GitHub [6.7302091035327285]
  Large language models (LLMs) are increasingly being integrated into software development processes.<n>The ability to generate code and submit pull requests with minimal human intervention, through the use of autonomous AI agents, is poised to become a standard practice.<n>We empirically study 567 GitHub pull requests (PRs) generated using Claude Code, an agentic coding tool, across 157 open-source projects.
  arXiv  Detail & Related papers  (2025-09-18T08:48:32Z)
• R&D-Agent: An LLM-Agent Framework Towards Autonomous Data Science [70.1638335489284]
  High-level machine learning engineering tasks remain labor-intensive and iterative.<n>We introduce R&D-Agent, a comprehensive, decoupled, and framework that formalizes the machine learning process.<n>R&D-Agent defines the MLE into two phases and six components, turning agent design for MLE into a principled, testable process.
  arXiv  Detail & Related papers  (2025-05-20T06:07:00Z)
• When Disagreements Elicit Robustness: Investigating Self-Repair Capabilities under LLM Multi-Agent Disagreements [56.29265568399648]
  We argue that disagreements prevent premature consensus and expand the explored solution space.<n>Disagreements on task-critical steps can derail collaboration depending on the topology of solution paths.
  arXiv  Detail & Related papers  (2025-02-21T02:24:43Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.