Vývoj pluginu pro Visual Studio Code za pomoci AI

This bachelor thesis deals with the development of an educational software tool for source code visualization and concurrently examines the process of application development with the close assistance of artificial intelligence agents. The main objective is to design and implement an extension for the Visual Studio Code editor capable of generating UML class diagrams and kopenograms from source code in Java, Python, and C#. The second equal objective is to systematically document the iterative development of this tool with the assistance of the large language model Claude Opus 4.6 and to critically evaluate the benefits and pitfalls of this modern engineering approach. The resulting product is built on a three-tier architecture using the .NET 8 platform for code analysis and the React library in conjunction with the maxGraph tool for the interactive user interface. An analysis of the programming process revealed that the use of artificial intelligence radically accelerates prototyping and helps the developer efficiently overcome barriers when working with unknown technologies. However, the thesis also identifies fundamental limits of this collaboration, which include the tendency of the model toward unnecessary code complexity, occasional looping during bug fixes, and the ability to persuasively defend suboptimal architectural decisions. The conclusions confirm that artificial intelligence transforms the role of the programmer into the position of a software architect, while achieving a high-quality result requires setting a strict communication methodology and maintaining critical human oversight.