Towards LLM-Assisted System Testing for Microservices

As modern applications are being designed in a distributed, Microservices Architecture (MSA), it becomes increasingly difficult to debug and test those systems. Typically, it is the role of software testing engineers or Quality Assurance (QA) engineers to write software tests to ensure the reliability of applications, but such a task can be labor-intensive and time-consuming. In this paper, we explore the potential of Large Language Models (LLMs) in assisting software engineers in generating test cases for software systems, with a particular focus on performing end-to-end (black-box) system testing on web-based MSA applications. We present our experience building Kashef, a software testing tool that utilizes the advanced capabilities of current LLMs in code generation and reasoning, and builds on top of the concept of communicative agents.