Unit testing, the bedrock of software development, is where each code unit is rigorously checked for correctness. Traditionally, crafting these tests has been time-consuming and demanding, a manual process ripe for automation. Large Language Models (LLMs) have shown promise in automating code generation, but previous attempts have mainly focused on small, isolated code snippets rather than entire classes—a more realistic scenario. Researchers have now developed AGONETEST, a fully automated system that generates and evaluates test suites for Java projects. This system leverages a new dataset, CLASSES2TEST, built upon the existing METHODS2TEST dataset, to generate test suites at the class level. By cloning repositories, saving commit hashes, and mapping test classes, the dataset provides a realistic testing environment, allowing test suites to grow in complexity. AGONETEST's magic comes from its automation. The system smartly pulls contextual information from the project's setup, such as the Java version and testing framework used, to guide the LLM. It integrates seamlessly with libraries like JaCoCo for coverage, PiTest for mutation testing (checking how tests react to code changes), and TsDetect to identify code smells that can make tests harder to maintain. The results? Promising, yet with room to grow. While AGONETEST efficiently produces class-level tests, addressing a key limitation of previous research, the initial tests showed that some generated code still needed fixes. In tests using GPT-4 and GPT-3.5 turbo, around 75% of the test classes compiled without errors, and about a third produced successful, error-free tests. This suggests that while LLMs excel at creating tests, they sometimes need a bit of refinement, with future work targeting automated fixes for common code issues and enhancing their performance in identifying potential bugs. This system represents a significant step towards fully automated testing, offering a valuable tool for developers seeking to enhance the effectiveness and efficiency of their testing process.