Requirement-Grounded LLM Agents for Functional Testing of React Applications: A Controlled Study Protocol, Benchmark Specification, and Artifact Audit

This work presents a controlled study protocol, benchmark specification, and artifact audit for investigating requirement-grounded large language model (LLM) agents for functional testing of React applications. The study design focuses on requirement-to-test traceability, LLM-assisted functional test generation, browser-based testing, Playwright-oriented test execution, mutation testing, test-oracle considerations, defect detection, and reproducibility. The manuscript defines the proposed methodology, benchmark structure, evaluation design, and artifact requirements for a controlled empirical study. This upload represents the protocol and benchmark specification stage and does not report completed experimental results.