Explainable AI Agents for Autonomous System Evaluation: Integrating SHAP-Based Decision Attribution With Hierarchical Planning in Large Language Models. JAAIR 2026, 5 (1).