RAG-Reflect: Agentic Retrieval-Augmented Generation with Reflections for Comment-Driven Code Maintenance on Stack Overflow

TL;DR

This paper introduces RAG-Reflect, a modular framework using retrieval, reasoning, and self-reflection with large language models to predict whether user comments on Stack Overflow trigger code edits, achieving high accuracy without task-specific training.

Contribution

The paper presents RAG-Reflect, a novel, task-agnostic framework that combines retrieval-augmented reasoning and self-reflection to predict comment-edit causality in software maintenance.

Findings

RAG-Reflect achieves an F1 score of 0.78, outperforming traditional baselines.

The framework approaches fine-tuned model performance without retraining.

Both retrieval and reflection modules significantly improve prediction accuracy.

Abstract

User comments on online programming platforms such as Stack Overflow play a vital role in maintaining the correctness and relevance of shared code examples. However, the majority of comments express gratitude or clarification, while only a small fraction highlight actionable issues that drive meaningful edits. This paper demonstrates how agentic AI principles can revolutionize software maintenance tasks by presenting RAG-Reflect, a modular framework that achieves fine-tuned-level performance for valid comment-edit prediction without task-specific training. Valid Comment-Edit Prediction (VCP) is the task of determining whether a user comment directly triggered a subsequent code edit. The framework integrates large language models (LLMs) with retrieval-augmented reasoning and self-reflection mechanisms. RAG-Reflect operates through a three-stage runtime workflow built on a one-time pattern analysis phase. During initialization, an Interpretation module analyzes the knowledge base to generate validation rules. At inference time, the system (1) retrieves contextual examples, (2) reasons about comment-edit causality, and (3) reflects on decisions using the pre-established rules. We evaluate RAG-Reflect on the publicly available SOUP benchmark, achieving Precision = 0.81, Recall = 0.74, and F1 = 0.78, outperforming traditional baselines (e.g., Logistic Regression, XGBoost, different prompting techniques) and closely approaching the performance of fine-tuned models (F1 = 0.773) without retraining. Our ablation and stage-level analyses show that both retrieval and reflection modules substantially enhance performance.