Web Application for the Automatic Code Generation of Fullstack Projects Using Generative Artificial Intelligence: GPT-5

David-Polanco A.; Prado-Vargas M.; Wong L. · 2026 8th International Conference on Software Engineering and Computer Science, CSECS 2026 - Conference Proceedings (2026) · DOI: 10.1109/csecs69124.2026.11541607

Resumen

Due to the growing interest in the automatic generation of web applications using Generative Artificial Intelligence (GenAI), several tools have emerged that promise to reduce development time and improve productivity. In practice, current GenAI- based tools are often limited to frontend generation or restricted technology stacks, which leads to technical inconsistencies between layers, rework, and the lack of integrated quality, security, and accessibility controls. Therefore, this study proposes a web application that, starting from a Design System, generates fullstack projects through the orchestration of GenAI. The model is organized into four phases: (i) analysis of Design System tokens, (ii) GenAI orchestration, (iii) generation of the frontend in Angular and the backend in.NET Core, and (iv) quality and accessibility validation with SonarQube, Lighthouse and WAVE, additionally integrating automatic API documentation and security based on JWT/OAuth2. For validation, a functional prototype was implemented and experiments were conducted with two students, who generated multiple Angular/.NET modules and obtained automatic metrics of bugs, vulnerabilities, code smells, performance and accessibility. The results show that the solution successfully completes the end-to-end flow, produces projects aligned with the Design System and provides objective quality indicators that facilitate inspection and improvement, positioning it as a viable alternative for academic and, potentially, professional contexts.

Instituciones: Universidad Peruana de Ciencias Aplicadas