Software test cases are the most important practice for software testing to establish confidence in business, it is importance for controlling testing processes, team developers and product owner may do it poorly, because they don't have the ability of understanding how to use test cases properly, and why it is necessary they also may not have framework to generate test cases automatically. In this work, an approach is proposed to help agile team and stakeholders to generate legitimate test cases automatically from their written English requirements using unified modeling language models, natural language processing used for analyzing the requirements and assigning each token to an appropriate part-of-speech, and identified rules should be applied in order to create extensible markup language schema format suitable to generate unified modeling language class diagrams. Whereas, relationships between classes are identified based on proposed algorithm. This work aiming to clarify possible links between the requirements. Requirements repository is used for mapping extensible markup language tags to build class diagram, test cases will be generated from classes diagram to overcome the weakness of providing test cases manually in agile approach.