ArchAgent: Scalable Legacy Software Architecture Recovery with LLMs

2026-01-19

AI Summary

This work addresses the challenge of accurately recovering the architecture of large-scale legacy software systems, which is hindered by architectural drift, missing relationships, and the context length limitations of large language models (LLMs). To overcome these issues, the authors propose a scalable, agent-based framework that integrates static analysis, adaptive code segmentation, and LLM-driven synthesis. The framework employs a context pruning mechanism to generate multi-perspective, business-aligned architectural views and leverages cross-repository dependency analysis to identify critical business modules. Experimental evaluation on representative large GitHub projects demonstrates that the proposed approach significantly outperforms existing baselines. Ablation studies confirm the positive impact of cross-repository context on architectural accuracy, and real-world case studies successfully reconstruct core business logic from legacy systems.

Technology Category

• Cognitive Modeling & Cognitive Systems: Agent Architectures
• Multiagent Systems: Agent/AI Theories and Architectures
• Natural Language Processing: Code Generation / Program Synthesis from Natural Language

Application Category

• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Search and Retrieval-Augmented AI: Agentic search

Abstract

Recovering accurate architecture from large-scale legacy software is hindered by architectural drift, missing relations, and the limited context of Large Language Models (LLMs). We present ArchAgent, a scalable agent-based framework that combines static analysis, adaptive code segmentation, and LLM-powered synthesis to reconstruct multiview, business-aligned architectures from cross-repository codebases. ArchAgent introduces scalable diagram generation with contextual pruning and integrates cross-repository data to identify business-critical modules. Evaluations of typical large-scale GitHub projects show significant improvements over existing benchmarks. An ablation study confirms that dependency context improves the accuracy of generated architectures of production-level repositories, and a real-world case study demonstrates effective recovery of critical business logics from legacy projects. The dataset is available at https://github.com/panrusheng/arch-eval-benchmark.

Problem

• legacy software
• architecture recovery
• architectural drift
• Large Language Models
• software architecture

Innovation

• Architecture Recovery
• Large Language Models
• Agent-based Framework
• Cross-repository Analysis
• Legacy Software