Agile development, characterized by rapid delivery and continuous iteration, has become the mainstream paradigm in modern software development. However, ensuring the reliability of systems on a large scale remains a significant challenge within this framework. The Agile-SOFL methodology addresses this gap by integrating the flexibility of Agile with the rigor of formal methods, providing a theoretical foundation for structured and verifiable software development. The rise of Large Language Models (LLMs) presents an opportunity to explore the profound potential of combining classic software engineering theory with modern AI technology. To this end, we introduce a novel software tool that leverages a key component of Agile-SOFL, namely the Functional Scenario Form (FSF), as a semi-formal bridge between user requirements and LLM-based code generation. The approach employs LLMs to automate the code generation phase of the Agile-SOFL process, significantly boosting development efficiency. At the same time, it utilizes the structured nature of FSF specifications to constrain and guide the generative process of LLMs, thereby ensuring the accuracy and reliability of the resulting code. We present the design and implementation of this tool and demonstrate its practical utility through a case study.