Official checkpoints for VP-VLA: Visual Prompting as an Interface for Vision-Language-Action Models