Return to Article Details Quantifying Correction Recovery and Robustness Under Prompt Perturbation in Language Model Agents Download Download PDF