Abstract: Producing executable code from natural-language directives via Large Language Models (LLMs) involves obstacles like semantic uncertainty and the requirement for task-focused context ...