Achieving high-quality software is a primary goal in software engineering, which requires rigorous validation and verification processes throughout testing activities. Although manual testing can be effective, it is often time-consuming and resource-intensive, leading to an increased demand for automated solutions. Recent advances in large-language models (LLMs) have significantly influenced various domains within software engineering, including requirements analysis, test automation, and debugging. This article explores an agent-oriented frame-work for automated software testing that leverages the capabilities of LLMs to minimize human intervention and improve testing efficiency. The proposed approach integrates LLMs to automate the generation of unit tests, visualization of call graphs, and execution and reporting of tests. The framework is evaluated using multiple applications developed in Python and Java, demonstrating high test coverage and operational efficiency. The findings of the article emphasize the potential of LLM-powered agents to streamline software testing workflows and address critical challenges related to scalability and accuracy.