一句话：提出统一软件工程智能体 USEagent 与 USEbench 基准，使 LLM 智能体跨编码、测试与修复等多任务协同。
软件工程不止于编码，还涉及项目的维护与演化，但已有 LLM 智能体多为测试、调试、修复等单一任务定制。本文提出统一软件工程智能体 USEagent：以 LLM 为推理引擎自主调用外部工具，统一编排编码、测试、补丁修复等多种能力，以应对修复不完整补丁、添加新特性、接手他人代码等复杂场景。为此构建 USEbench，从 SWE-bench、SWT-bench、REPOCOD 等基准混合出覆盖编码、测试与打补丁的 1271 个仓库级任务。结果显示 USEagent 优于 OpenHands CodeActAgent 等通用智能体，在 SWE-bench 的 issue 解决上达到与 AutoCodeRover 等维护专用智能体相当的水平，且适用面更广；同时暴露出其在部分编码任务上的能力缺口。
The growth of Large Language Model (LLM) technology has raised expectations for automated coding. However, software engineering is more than coding and is concerned with activities including maintenance and evolution of a project. In this context, the concept of LLM agents has gained traction, which utilize LLMs as reasoning engines to invoke external tools autonomously. But is an LLM agent the same as an AI software engineer? In this paper, we seek to understand this question by developing a Unified Software Engineering agent or USEagent. Unlike existing work which builds specialized agents for specific software tasks such as testing, debugging, and repair, our goal is to build a unified agent which can orchestrate and handle multiple capabilities. This gives the agent the promise of handling complex scenarios in software development such as fixing an incomplete patch, adding new features, or taking over code written by others. We envision USEagent as the first draft of a future AI Software Engineer which can be a team member in future software development teams involving both AI and humans. To evaluate the efficacy of USEagent, we build a Unified Software Engineering bench (USEbench) comprising of myriad tasks such as coding, testing, and patching. USEbench is a judicious mixture of tasks from existing benchmarks such as SWE-bench, SWT-bench, and REPOCOD. In an evaluation on USEbench consisting of 1,271 repository-level software engineering tasks, USEagent shows improved efficacy compared to existing general agents such as OpenHands CodeActAgent. For specific tasks such as issue resolution in SWE-bench, USEagent demonstrates efficacy comparable to agents such as AutoCodeRover (which are focused on software maintenance), while maintaining applicability to a broader range of software engineering tasks. There exist gaps in the capabilities of USEagent for certain coding tasks, which provides hints on further developing the AI Software Engineer of the future.
@inproceedings{abs-2506-14683,
title = {Unified Software Engineering Agent as AI Software Engineer},
author = {Leonhard Applis and Yuntong Zhang and Shanchao Liang and Nan Jiang and Lin Tan and Abhik Roychoudhury},
booktitle = {ICSE},
year = {2026},
doi = {10.1145/3744916.3773202}
}