Improving Automated Issue Resolution via Comprehensive Repository Exploration

YINGWEI MA, Yue Liu

Track: long paper (up to 9 pages)

Keywords: Automatic Software Engineering (ASE), Software Engineering Agents, Monte Carlo Tree Search

TL;DR: LingmaAgent is an AI-powered software engineering tool that comprehensively analyzes entire code repositories to effectively resolve issues, outperforming existing methods in both benchmarks and real-world industrial applications.

Abstract: This paper presents LingmaAgent, a novel Automated Software Engineering method designed to comprehensively understand and utilize whole software repositories for issue resolution. LingmaAgent addresses the limitations of existing LLM-based agents that primarily focus on local code information. Our approach introduces a top-down method to condense critical repository information into a knowledge graph, reducing complexity, and employs a Monte Carlo tree search based strategy enabling agents to explore entire repositories. We guide agents to summarize, analyze, and plan using repository-level knowledge, allowing them to dynamically acquire information and generate patches for real-world GitHub issues. In extensive experiments, LingmaAgent demonstrated significant improvements, achieving an 18.5\% relative improvement on the SWE-bench Lite benchmark compared to SWE-agent. In production deployment and evaluation at a major cloud computing industrial partner, LingmaAgent automatically resolved 16.9\% of in-house issues faced by development engineers, and solved 43.3\% of problems after manual intervention. Additionally, we have open-sourced a Python prototype of LingmaAgent for reference by other industrial developers~\footnote{\url{https://anonymous.4open.science/r/RepoU-4E43/README.md}}.

Anonymization: This submission has been anonymized for double-blind review via the removal of identifying information such as names, affiliations, and identifying URLs.

Submission Number: 14