Path-Aware Defense Against Prompt Injection Attacks in Retrieval-Augmented Large Language Models. JAAIR. 2026;1(1). Accessed July 29, 2026. https://www.jaair.org/index.php/home/article/view/132