Humans in Control: A Methodological Framework for Quality Assurance in Agentic Software Engineering

• Matheus Silveira USP
• Paulo Meirelles USP
• Igor Steinmacher Northern Arizona University
• Fabio Kon USP

DOI: https://doi.org/10.5753/se4as.2026.32407

Resumo

As Large Language Models (LLMs) continue to advance, their application in software engineering has expanded considerably, particularly in automated code and test generation, which significantly accelerates the development process. However, the resulting increase in code output often leaves developers overwhelmed by the volume of code requiring review, often leading to a progressive degradation in code quality and the accumulation of cognitive debt among development teams as they struggle to maintain a comprehensive understanding of an increasingly complex codebase. To address the high volume of AI-generated code, this paper introduces a multi-agent methodology for quality assurance in agentic software engineering, specifically aimed at resolving code smells and other maintainability issues reported by static analysis tools, rather than targeting feature generation or software evolution in general. The framework establishes a collaborative workflow that preserves software integrity while keeping humans firmly in control of the engineering process, and it is being supported by an under development dedicated software tool.

Palavras-chave: Code Refactoring, LLM, Agentic Software Engineering, Multi-agent

Referências

Yuanfang Cai, Lanting He, Jun Qian, Yony Kochinski, Nan Zhang, Ciera Jaspan, andAntonio Bianco. 2025. UnderstandingArchitectural Complexity,Maintenance Burden, and Developer Sentiment – A Large-Scale Study. In Proceedings of the IEEE/ACM 47th International Conference on Software Engineering (ICSE ’25). IEEE, Ottawa, ON, Canada, 2176–2187. DOI: 10.1109/ICSE55347.2025.00168

GitLab and TheHarris Poll. 2026. Global DevSecOps Report: The Intelligent Software Development Era: How AI will redefine DevSecOps in 2026 and beyond. Report. GitLab. [link] Survey conducted from July 31, 2025, to August 15, 2025, among 3,266 DevSecOps professionals..

Kyogo Horikawa, Kosei Horikawa, Yutaro Kashiwa, Hidetake Uwano, and Hajimu Iida. 2026. Do AI Agents Really Improve Code Readability?. In Proceedings of the 2026 IEEE/ACM 23rd International Conference on Mining Software Repositories (MSR ’26). ACM, Rio de Janeiro, Brazil. MSR 2026 Mining Challenge Track.

James Ivers, Robert L. Nord, Ipek Ozkaya, Chris Seifried, Christopher S. Timperley, andMarouane Kessentini. 2022. Industry experiences with large-scale refactoring. In Proceedings of the 30th ACM Joint European Software Engineering Conference and Symposium on the Foundations of Software Engineering (ESEC/FSE ’22). ACM, New York, NY, USA, 1544–1554. DOI: 10.1145/3540250.3558954

Gabriel Orlanski, Devjeet Roy, Alexander Yun, Changho Shin, Alex Gu, Albert Ge, Dyah Adila, Nicholas Roberts, Frederic Sala, and Aws Albarghouthi. 2026. SlopCodeBench: BenchmarkingHowCodingAgents DegradeOver Long-Horizon Iterative Tasks. arXiv preprint arXiv:2603.24755 (2026).

Lukas Ottenhof, Daniel Penner, Abram Hindle, and Thibaud Lutellier. 2026. How Do Agents Refactor: An Empirical Study. In Proceedings of the 2026 IEEE/ACM 23rd International Conference on Mining Software Repositories (MSR ’26). ACM, Rio de Janeiro, Brazil. MSR 2026 Mining Challenge Track.

Khouloud Oueslati, Maxime Lamothe, and Foutse Khomh. 2026. RefAgent: A Multi-Agent LLM-Based Framework for Automatic Software Refactoring. In Proceedings of the 2026 IEEE/ACM 48th International Conference on Software Engineering (ICSE ’26). ACM, Rio de Janeiro, Brazil, 92–104. DOI: 10.1145/3744916.3773153

Wannita Takerngsaksiri, Jirat Pasuksmit, Patanamon Thongtanunam, Chakkrit Tantithamthavorn, Ruixiong Zhang, Fan Jiang, Jing Li, Evan Cook, Kun Chen, and Ming Wu. 2025. Human-In-The-Loop Software Development Agents. In Proceedings of the 2025 IEEE/ACM 47th International Conference on Software Engineering: Software Engineering in Practice (ICSE-SEIP ’25). IEEE, Ottawa, ON, Canada, 342–352. DOI: 10.1109/ICSE-SEIP66354.2025.00036

Kla Tantithamthavorn, Yaotian Zou, Andy Wong, Michael Gupta, Zhe Wang, Mike Buller, Ryan Jiang, Matthew Watson, Minwoo Jeong, Kun Chen, and Ming Wu. 2026. RovoDev Code Reviewer: A Large-Scale Online Evaluation of LLM-Based Code Review Automation at Atlassian. In Proceedings of the 48th IEEE/ACM International Conference on Software Engineering: Software Engineering in Practice (ICSE-SEIP ’26). ACM, Rio de Janeiro, Brazil, 34–45. DOI: 10.1145/3786583.3786851

Priyan Vaithilingam, Tianyi Zhang, and Elena Glassman. 2022. Expectation vs. Experience: Evaluating the Usability of Code Generation Tools Powered by Large Language Models. In Proceedings of the CHI Conference on Human Factors in Computing Systems Extended Abstracts (CHI EA ’22). ACM, New Orleans, LA, USA, 1–7. DOI: 10.1145/3491101.3519665

Nalin Wadhwa, Jui Pradhan, Atharv Sonwane, Surya Prakash Sahu, Nagarajan Natarajan, Aditya Kanade, Suresh Parthasarathy, and Sriram Rajamani. 2024. CORE: Resolving Code Quality Issues using LLMs. Proceedings of the ACM on Software Engineering 1, FSE, Article 36 (2024), 23 pages. DOI: 10.1145/3643762