SWE-Adept: An LLM-Based Agentic Framework for Deep Codebase Analysis and Structured Issue Resolution

2026-03-01

AI Summary

This work addresses key challenges faced by large language models in warehouse-scale software engineering tasks, including difficulties in codebase navigation, insufficient context management, and the absence of systematic repair mechanisms. To overcome these limitations, the authors propose SWE-Adept, a dual-agent collaborative framework. A localization agent employs proxy-guided depth-first search to precisely identify fault locations, while a repair agent executes modifications through adaptive planning and structured solving strategies. The framework integrates Git and leverages a shared working memory with execution indexing to enable traceable and verifiable repair workflows. It further supports branch exploration and failure rollback, achieving up to a 4.7% improvement in end-to-end resolution rates on both SWE-Bench Lite and Pro benchmarks, significantly outperforming existing approaches.

Technology Category

• Planning, Routing, and Scheduling: Replanning and Plan Repair
• Search and Optimization: Sampling/Simulation-based Search
• Multiagent Systems: Teamwork

Application Category

• Search and Retrieval-Augmented AI: Agentic search
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Large language models (LLMs) exhibit strong performance on self-contained programming tasks. However, they still struggle with repository-level software engineering (SWE), which demands (1) deep codebase navigation with effective context management for accurate localization, and (2) systematic approaches for iterative, test-driven code modification to resolve issues. To address these challenges, we propose SWE-Adept, an LLM-based two-agent framework where a localization agent identifies issue-relevant code locations and a resolution agent implements the corresponding fixes. For issue localization, we introduce agent-directed depth-first search that selectively traverses code dependencies. This minimizes issue-irrelevant content in the agent's context window and improves localization accuracy. For issue resolution, we employ adaptive planning and structured problem solving. We equip the agent with specialized tools for progress tracking and Git-based version control. These tools interface with a shared working memory that stores code-state checkpoints indexed by execution steps, facilitating precise checkpoint retrieval. This design enables reliable agent-driven version-control operations for systematic issue resolution, including branching to explore alternative solutions and reverting failed edits. Experiments on SWE-Bench Lite and SWE-Bench Pro demonstrate that SWE-Adept consistently outperforms prior approaches in both issue localization and resolution, improving the end-to-end resolve rate by up to 4.7%.

Problem

• repository-level software engineering
• codebase navigation
• issue localization
• test-driven code modification
• context management

Innovation

• agent-directed depth-first search
• structured problem solving
• shared working memory
• Git-based version control
• two-agent framework