It is a pure front-end tool for testing the security boundaries of large language models, helping researchers to find and fix potential security vulnerabilities and improve the security and reliability of AI systems.