Unit Test Generation Multi-Agent AI System for Enhancing Software Documentation and Code Coverage

Unit Test Generation Multi-Agent AI System for Enhancing Software Documentation and Code Coverage

연구 분야: Software Development

학회: 2024 32nd Telecommunications Forum (TELFOR)

Software development necessitates a robust testing plan though test development can be laborious and nonappealing task. We explore the utilization of the application artificial intelligence agents for generating and executing unit tests, enhancing the “Mostly Basic Python Problems” dataset. We employ behavior-driven development within a three-agent system to generate user stories and unit tests. Empirical results indicate improvements in branch coverage, illustrating the effective utilization of large language models in software testing and development processes.