arXiv cs.AI· Mohammad Wathiq Soualhi, Basem Suleiman·· 6 天前AI 评分34
AI 智能体 PR 同时修改应用代码与部署基础设施的因果研究
Co-Modifying Application Code and Deployment Infrastructure: A Causal Study of AI Agent Pull Requests
AI 导读
一项因果研究分析了 AIDev v5 和 MOSAIC-3M 两个数据集中的 26,999 个 PR，发现同时修改应用代码与部署基础设施的双域 PR 仅占智能体 PR 的 4.48%–4.55%，却呈现双峰式集成结果。
正文 · 原文
Abstract:Current software engineering agents are evaluated predominantly inside isolated sandboxes against synchronous local tests. However, deploying features in collaborative repositories requires modifying continuous integration (CI/CD) pipelines, containers, and provisioning manifests alongside application logic. Because agents lack deployment awareness, attempting cross-domain modifications introduces unverified dependencies into asynchronous integration pipelines. To isolate the causal integration overhead of dual-scope pull requests (PRs)-modifying both application code and deployment infrastructure-we evaluated 26,999 PRs across two large datasets (AIDev v5 and MOSAIC-3M) encompassing six agent architectures and an aligned human baseline. Using a deterministic four-tier file taxonomy and an exact 4D strata-matching framework, we isolated integration metrics under conditional exchangeability while controlling for application volume. We find agents operate primarily as application coders or chore workers, with maintenance tasks capturing 17.70% to 25.41% of agent PRs. Dual-scope modifications represent only 4.48% to 4.55% of agentic PRs, yet trigger bimodal integration outcomes. In the first mode, explicitly reviewed contributions experience protracted delays, adding a median penalty of +13.13 hours and multiplying time-to-merge by 3.24x, driven mechanically by CI/CD and provisioning manifests. In the second mode, maintainer gatekeeping exhibits a structural bypass: even for external contributors lacking repository write access, agent identity reduces the odds of human review by 63.7% compared to human peers. Consequently, 43.1% to 56.2% of dual-scope PRs bypass inspection to merge silently in a median of under 31 minutes. These uninspected modifications compromise build verification, demonstrating the need for deployment-aware benchmarks and strict path-based access controls.
| Comments: | 21 pages, 2 figures |
| Subjects: | Software Engineering (cs.SE); Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2610.12849 [cs.SE] |
| (or arXiv:2610.12849v1 [cs.SE] for this version) | |
| https://doi.org/10.48550/arXiv.2610.12849 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Mohammad Wathiq Soualhi [view email]
[v1]
Fri, 9 Oct 2026 01:16:04 UTC (131 KB)
来源：arXiv cs.AI · arxiv.org