Reinforcement Learning-Guided Prompt Insertion for Robust Language Model Alignment. JDIAS 2026, 1 (1).