# Auriga Technical Report: A Task-Agnostic Agent Harness for Zero-Shot Embodied Manipulation > Source: > Published: 2026-11-10 14:30:14+00:00 Vision-language models can interpret instructions and plan manipulation tasks, yet struggle when acting directly as robot policies. On RoboDojo, a large planning-execution gap separates high-level task decisions from physical execution: removing arm execution raises GPT-6 Astra's success rate from 2