Sequential reasoning in agent systems has been significantly advanced by large language
models (LLMs), yet existing approaches face limitations. Reflection-driven reasoning relies …