cs.AIJun 19, 2026

AutoRAS: Learning Robust Agentic Systems with Primitive Representations

Authors: Yang YueXuancheng ZhuYuyang MaGuoshun NanZihan DouJingru ShanCongyu GuoJi Zhang+2 more

Abstract

The automated design of agentic systems offers a promising pathway for scaling large language models (LLMs) beyond single-agent reasoning. While prior work has advanced task performance through handcrafted or automatically generated multi-agent workflows, robustness is often treated as an afterthought, leaving systems vulnerable to external adversaries and internal failures. We propose AutoRAS, a framework for the Automated design of Robust Agentic Systems. AutoRAS formulates system design as generating a sequence of symbolic primitives that jointly encode structural connectivity and behavioral actions, and learns to optimize this sequence using execution-derived safety signals and flow-based sequence-level objectives. Extensive experiments show that AutoRAS achieves the best performance in both vanilla and adversarial settings, with the smallest performance degradation under attacks. Further analyses demonstrate strong transferability, stable optimization behavior, stability across primitive sets, and favorable cost trade-offs. Our code is available at \href\href{https://github.com/guohezuy/AutoRAS}{\text{this https URL}}.

Explore similar work

CardsList
  1. Learning to Construct Practical Agentic Systems

    May 29, 2026Aditya Kumar, Zhihan Lei, Jerry Yan +6Agentic SystemsInference Cost

  2. Agents in the Wild: Where Research Meets Deployment

    Jul 21, 2026Grace Hui Yang, Pranav N. Venkit, Hooman Sedghamiz +3Agentic SystemsVerifier