cs.AINov 3, 2026

DoGNAVY-Exploit: Harnessing Parallel Open-Weight Agents for Stage-Aware Vulnerability Exploitation

Authors: Yuxuan Wang, Dadi Guo, Weijie Guo, Zikai Xu, Chen Lu, Yizhe Zhuang, Peng Huang, Yuejin Xie, 

Organizations: Shanghai Artificial Intelligence Laboratory · HKUST · DARKNAVY · Tsinghua University

Abstract

Frontier LLM agents can turn known vulnerabilities into working exploits, but are costly and often access-restricted due to their potential for malicious use. Open-weight models offer a more accessible alternative, but struggle to maintain progress across long exploitation chains. They also often fail to apply relevant security knowledge at the appropriate stage and in the right context to identify and evaluate promising attack directions. To address these limitations, we propose DoGNAVY-Exploit, a multi-agent harness built around two complementary designs: stage-aware decomposition with progressive expert knowledge disclosure, and hierarchical orchestration with parallel exploration. Specifically, DoGNAVY-Exploit uses configurable milestones to recast long-horizon exploitation as short, verifiable capability transitions and provides only the knowledge relevant to each stage. Within each milestone, a hierarchical architecture separates strategic planning from parallel execution: a planner proposes diverse attack routes, executors test them concurrently, and a controller accepts progress only when supported by runtime evidence. A shared structured memory consolidates verified facts, artifacts, replay steps, and failed routes across branches, allowing agents to build on collective progress and carry validated knowledge into subsequent milestones. In controlled benchmark environments, DoGNAVY-Exploit with GLM-5.2 achieves 97.5% Pass@1 on Cybench, and 97.5% zero-day and 100% one-day Pass@1 on CVE-Bench. On ExploitGym, it solves 79 tasks with GLM-5.2 and 157 with GLM-5.3, while the corresponding public Claude Code baselines solve 39 and 130 tasks, respectively. Ablation results further demonstrate that the stage-aware orchestration, expert knowledge, parallel exploration, and memory designs can substantially improve the exploitation capabilities of open-weight agents.

Figures & tables

Appendix figures & tables1 asset

Supplementary material from the paper’s appendix.

Appendix

Explore similar work

CardsList
  1. ExploitGym: Can AI Agents Turn Security Vulnerabilities into Real Attacks?

    May 11, 2026Zhun Wang, Nico Schiller, Hongwei Li +13AI Agent SecurityAI Agent Security Benchmarks

  2. Decoupling Reconnaissance and Exploitation: Measuring the Capability Boundaries of LLM-Based Web Penetration Testing

    Jun 24, 2026Liwei Yu, Shuo Li, Ming Zhou +2LLM Agent EvaluationAI Agent Security Benchmarks