两篇新研究论文分析了AI代理在软件开发中的有效性，特别关注拉取请求。第一篇论文《迈向指令即代码》发现，虽然指令文件可以指导GitHub Copilot等AI代理，但它们对拉取请求成功率的影响喜忧参半，一些项目有所改善，另一些项目则出现下降。第二篇论文《理解代理拉取请求修复的拒绝原因》调查了AI生成的修复被拒绝的原因，确定了不正确的实现、CI管道失败和代理限制是主要原因，近一半的提议修复被丢弃。 AI
影响 研究强调了AI代理集成中的挑战，表明需要更好的指导和任务优先级排序，以提高软件开发的效率并减少浪费的精力。
排序理由 两篇发表在arXiv上的学术论文，分析了AI代理在软件工程中的表现。
AI 生成摘要 · Google Gemini · 来自 4 个来源。 我们如何撰写摘要 →
arXiv:2606.13449v1 Announce Type: cross Abstract: AI-agents (e.g., GitHub Copilot) collaborate as teammates in different software engineering tasks, including code generation proposed through pull requests (Agentic-PRs). For better agent efficiency, developers create instruction …
arXiv:2606.13468v1 Announce Type: cross Abstract: AI coding agents are increasingly used to generate pull requests (PRs) that propose code fixes in software projects. From a first exploration of the AIDev dataset, we find that 46.41\% of the fixes proposed by the agents Copilot, …
AI coding agents are increasingly used to generate pull requests (PRs) that propose code fixes in software projects. From a first exploration of the AIDev dataset, we find that 46.41\% of the fixes proposed by the agents Copilot, Devin, Cursor, and Claude are rejected. This repre…
AI-agents (e.g., GitHub Copilot) collaborate as teammates in different software engineering tasks, including code generation proposed through pull requests (Agentic-PRs). For better agent efficiency, developers create instruction files that guide the AI-agents, including how to n…