ReProAgent Tool-Augmented Multi-Stage Agentic Generation of Bug Reproduction Tests from Issue Reports

Jul 16, 2026

Upload by Eden

ReProAgent is a tool-augmented, multi-stage agent framework for generating executable bug reproduction tests directly from open-source issue reports. It addresses limitations of prompt-based pipelines by decomposing the workflow into four stages: bug localization, root cause analysis, test planning, and test generation. ReProAgent combines task-specific decomposition and reflection tools, retrieval from both textual sources and repository graphs, and runtime interaction with the execution environment. Experiments on SWT-bench-lite and SWT-bench-verified reach 58.43% and 70.30% reproduction success rates, outperforming baselines.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages23

Reading Time58 min

Introduction

Problem and Motivation

Proposed Approach

Experimental Evaluation

arXiv :2607 .09 123v 1 [ cs . SE] 10 Jul 2026 ReProAgent: Tool-Augmented Multi-Stage Agentic Generation of Bug Reproduction Tests from Issue Reports QUANJUN ZHANG, School of Computer Science and Engineering, Nanjing University of Science and Technology, China YI ZHENG and YE SHANG, State Key Laboratory for Novel Software Technology, Nanjing University Nanjing University, China WEIFENG SUN, Singapore Management University, Singapore HAICHUAN HU, School of Computer Science and Engineering, Nanjing University of Science and Technology, China CHUNRONG FANG and ZHENYU CHEN, State Key Laboratory for Novel Software Technology, Nanjing University, China LIANG XIAO, School of Computer Science and Engineering, Nanjing University of Science and Technology, China Reproduction tests help developers confirm reported issues and provide executable feedback for issue resolution, yet issue reports in open-source projects rarely include such tests. Recent studies have explored generating issue reproduction tests from issue reports with large language models, but existing approaches largely rely on prompt-based pipelines that retrieve textual context and generate tests. This limits their ability to understand how reported issues behave in repository-scale codebases and to flexibly organize the construction of reproduction tests. In this paper, we propose ReProAgent, a multi-stage agent framework for reproduction test generation from issue reports. Inspired by how developers manually reproduce reported issues, ReProAgent decomposes the task into four agen stages: bug localization, root cause analysis, test planning, and test generation. To support these stages, ReProAgent integrates task-specific tools for task decomposition and reflection, context retrieval from both textual sources and repository graphs, and runtime interaction with the execution environment. Experiments on SWT-bench-lite and SWT-bench-verified show that ReProAgent successfully reproduces 58.43% and 70.30% of issues, outperforming all baselines, with an average cost of $0.14 per instance. For example, when equipped with GPT-5-mini, ReProAgent exceeds OpenHands with the same backbone by 20.43 and 7.90 percentage points, respectively. ReProAgent also generalizes across multiple backbone LLMs and improves downstream issue resolution performance when integrated with existing repair approaches. CCS Concepts: • Software and its engineering → Software testing and debugging. Additional Key Words and Phrases: LLMs, Reproduction Test Generation, Agents, Knowledge Graph Authors’ Contact Information: Quanjun Zhang, [quanjunzhang@njust.edu.cn](quanjunzhang@njust.edu.cn), School of Computer Science and Engineering, Nanjing University of Science and Technology, Nanjing, China; Yi Zheng, [201250182@smail.nju.edu.cn](201250182@smail.nju.edu.cn); Ye Shang, yeshang@ [smail.nju.edu.cn](smail.nju.edu.cn), State Key Laboratory for Novel Software Technology, Nanjing University and Nanjing University, Nanjing, China; Weifeng Sun, [wfsun@smu.edu.sg](wfsun@smu.edu.sg), Singapore Management University, Singapore; Haichuan Hu, huhaichuan2024@ [gmail.com](gmail.com), School of Computer Science and Engineering, Nanjing University of Science and Technology, Nanjing, China; Chunrong Fang, [fangchunrong@nju.edu.cn](fangchunrong@nju.edu.cn); Zhenyu Chen, [zychen@nju.edu.cn](zychen@nju.edu.cn), State Key Laboratory for Novel Software Technology, Nanjing University, Nanjing, China; Liang Xiao, [xiaoliang@mail.njust.edu.cn](xiaoliang@mail.njust.edu.cn), School of Computer Science and Engineering, Nanjing University of Science and Technology, Nanjing, China. Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page. Copyrights for components of this work owned by others

FAQs

What problem does ReProAgent address in open-source issue fixing?

Issue reports commonly lack executable reproduction tests, making it harder for developers to confirm bugs before fixing and validate patches afterward. ReProAgent generates faithful reproduction tests from the issue text to reduce debugging and maintenance cost.

How does ReProAgent structure its test generation workflow?

ReProAgent decomposes the task into four agent stages: bug localization, root cause analysis, test planning, and test generation, mirroring how developers manually reproduce reported issues.

Which capabilities help ReProAgent generate better reproduction tests?

It integrates task-specific tools for decomposition and reflection, retrieves context from both textual sources and repository graphs, and interacts with the execution environment at runtime.

Jul 16, 2026

Upload by Eden