Source-Code-Aware RAG for Autonomous Technical Documentation Generation

Abstract

The primary objective of this project is to develop a system that autonomously generates software architecture documentation after analyzing source code from GitHub repositories using a Retrieval-Augmented Generation (RAG) architecture and Large Language Models (LLMs). The developed system features a single architecture comprising a user-friendly interface (frontend), a server architecture (backend), a vector-based database, and IEEE-compliant DOCX output. This project offers a solution that automates the documentation and reporting processes, which are a significant burden for developers. © 2026 IEEE.