Automating quality assurance for industrial web applications : a hybrid approach using API fuzzing and agentic UI testing
Wu, Yuewei (2025)
Julkaisun pysyvä osoite on
https://urn.fi/URN:NBN:fi:amk-2025120432662
https://urn.fi/URN:NBN:fi:amk-2025120432662
Tiivistelmä
This thesis evaluates automated testing strategies for comprehensive fault detection in specialized web applications, combining backend black-box API fuzzing with frontend AI-powered UI testing agents.
Conventional testing frameworks frequently struggle to handle intricate API dependencies and the volatility of dynamic user interfaces, leading to significant productivity bottlenecks in quality assurance. To address this gap, this thesis applies a hybrid automated approach capable of adapting to both backend complexity and frontend fluidity. Specifically, the study examines the efficacy of advanced fuzzing techniques for RESTful API validation alongside autonomous, agent-based architectures for UI verification.
The evaluation focuses on three quality dimensions, including autonomy, functional depth, and resilience. Experimental results demonstrate that while API fuzzing effectively uncovers hidden integration faults and security vulnerabilities, AI-driven UI agents significantly enhance resilience against interface changes compared to traditional scripting. Finally, this work combines these findings to deliver actionable recommendations for deploying hybrid automated testing pipelines. These insights provide a pathway for industrial domains, such as construction technology, to achieve faster, safer software releases with improved defect coverage.
Conventional testing frameworks frequently struggle to handle intricate API dependencies and the volatility of dynamic user interfaces, leading to significant productivity bottlenecks in quality assurance. To address this gap, this thesis applies a hybrid automated approach capable of adapting to both backend complexity and frontend fluidity. Specifically, the study examines the efficacy of advanced fuzzing techniques for RESTful API validation alongside autonomous, agent-based architectures for UI verification.
The evaluation focuses on three quality dimensions, including autonomy, functional depth, and resilience. Experimental results demonstrate that while API fuzzing effectively uncovers hidden integration faults and security vulnerabilities, AI-driven UI agents significantly enhance resilience against interface changes compared to traditional scripting. Finally, this work combines these findings to deliver actionable recommendations for deploying hybrid automated testing pipelines. These insights provide a pathway for industrial domains, such as construction technology, to achieve faster, safer software releases with improved defect coverage.