Meta-RAG on Large Codebases Using Code Summarization

2025-08-04

AI Summary

To address the low efficiency of defect localization in large-scale codebases, this paper proposes Meta-RAG, a multi-agent framework integrating code summarization, retrieval-augmented generation (RAG), and collaborative large language model (LLM) agents. Its core contributions are: (1) semantic compression of codebases via natural-language summaries, reducing raw code volume by 79.8% on average and yielding compact, structured knowledge representations; and (2) joint optimization of multi-granularity (file-level and function-level) retrieval and generation to enhance localization accuracy. Evaluated on the SWE-bench Lite benchmark, Meta-RAG achieves state-of-the-art performance with 84.67% file-level and 53.0% function-level defect identification accuracy. The framework bridges the gap between scalable code understanding and precise defect localization by synergizing summarization-driven knowledge distillation with RAG-enhanced reasoning across granularities.

Technology Category

• Search and Optimization: Metareasoning and Metaheuristics
• Natural Language Processing: Code Generation / Program Synthesis from Natural Language
• Machine Learning: Large Multimodal Models (LMMs)

Application Category

• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• 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 Model (LLM) systems have been at the forefront of applied Artificial Intelligence (AI) research in a multitude of domains. One such domain is software development, where researchers have pushed the automation of a number of code tasks through LLM agents. Software development is a complex ecosystem, that stretches far beyond code implementation and well into the realm of code maintenance. In this paper, we propose a multi-agent system to localize bugs in large pre-existing codebases using information retrieval and LLMs. Our system introduces a novel Retrieval Augmented Generation (RAG) approach, Meta-RAG, where we utilize summaries to condense codebases by an average of 79.8%, into a compact, structured, natural language representation. We then use an LLM agent to determine which parts of the codebase are critical for bug resolution, i.e. bug localization. We demonstrate the usefulness of Meta-RAG through evaluation with the SWE-bench Lite dataset. Meta-RAG scores 84.67 % and 53.0 % for file-level and function-level correct localization rates, respectively, achieving state-of-the-art performance.

Problem

• Localizing bugs in large pre-existing codebases efficiently
• Condensing codebases into compact natural language representations
• Improving bug localization accuracy using Meta-RAG approach

Innovation

• Meta-RAG for condensing codebases with summaries
• Multi-agent system for bug localization
• LLM agent identifies critical code parts