REAgent: Requirement-Driven LLM Agents for Software Issue Resolution

TL;DR

REAgent is a novel requirement-driven framework that enhances LLM-based software issue resolution by automatically creating and refining structured issue requirements, leading to significant improvements over existing methods.

Contribution

It introduces a structured, requirement-based approach for guiding LLMs in generating patches, addressing issues of input quality and ambiguity.

Findings

REAgent outperforms baselines with a 17.40% increase in resolved issues.

It automatically constructs and refines issue requirements to improve patch accuracy.

Experiments on three benchmarks validate the effectiveness of REAgent.

Abstract

Issue resolution aims to automatically generate patches from given issue descriptions and has attracted significant attention with the rapid advancement of large language models (LLMs). However, due to the complexity of software issues and codebases, LLM-generated patches often fail to resolve corresponding issues. Although various advanced techniques have been proposed with carefully designed tools and workflows, they typically treat issue descriptions as direct inputs and largely overlook their quality (e.g., missing critical context or containing ambiguous information), which hinders LLMs from accurate understanding and resolution. To address this limitation, we draw on principles from software requirements engineering and propose REAgent, a requirement-driven LLM agent framework that introduces issue-oriented requirements as structured task specifications to better guide patch generation. Specifically, REAgent automatically constructs structured and information-rich issue-oriented requirements, identifies low-quality requirements, and iteratively refines them to improve patch correctness. We conduct comprehensive experiments on three widely used benchmarks using two advanced LLMs, comparing against five representative or state-of-the-art baselines. The results demonstrate that REAgent consistently outperforms all baselines, achieving an average improvement of 17.40% in terms of the number of successfully-resolved issues (% Resolved).