This research introduces a novel two-step approach to automatically generate end-to-end (E2E) test code by leveraging product documentation and Large Language Models.

• Focuses on E2E testing rather than unit testing, addressing a significant gap in current research
• Creates test code from support materials like manuals, FAQs, and tutorials
• Uses tailored prompts to improve LLM performance in code generation tasks
• Demonstrates practical engineering applications for higher quality software testing

This approach matters for engineering teams by automating the creation of critical E2E tests that traditionally require significant effort, potentially reducing development time while maintaining robust quality assurance processes.

A Study on the Improvement of Code Generation Quality Using Large Language Models Leveraging Product Documentation