# Argus: Reorchestrating Static Analysis via a Multi-Agent Ensemble for Full-Chain Security Vulnerability Detection

Source: [Argus: Reorchestrating Static Analysis via a Multi-Agent Ensemble for Full-Chain Security Vulnerability Detection](<https://www.sota2.com/research/paper/69d732ae7d0172140ec76c4a>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2026-04-10T05:00:04.898Z

Authors: Zi Liang, Qipeng Xie, Jun He, Bohuan Xue, Weizheng Wang, Yuandao Cai, Fei Luo, Boxian Zhang, Haibo Hu, Kaishun Wu

Published: 2026-04-08T03:18:51.000Z

## Abstract

Recent advancements in Large Language Models (LLMs) have sparked interest in their application to Static Application Security Testing (SAST), primarily due to their superior contextual reasoning capabilities compared to traditional symbolic or rule-based methods. However, existing LLM-based approaches typically attempt to replace human experts directly without integrating effectively with existing SAST tools. This lack of integration results in ineffectiveness, including high rates of false positives, hallucinations, limited reasoning depth, and excessive token usage, making them impractical for industrial deployment. To overcome these limitations, we present a paradigm shift that reorchestrates the SAST workflow from current LLM-assisted structure to a new LLM-centered workflow. We introduce Argus (Agentic and Retrieval-Augmented Guarding System), the first multi-agent framework designed specifically for vulnerability detection. Argus incorporates three key novelties: comprehensive supply chain analysis, collaborative multi-agent workflows, and the integration of state-of-the-art techniques such as Retrieval-Augmented Generation (RAG) and ReAct to minimize hallucinations and enhance reasoning. Extensive empirical evaluation demonstrates that Argus significantly outperforms existing methods by detecting a higher volume of true vulnerabilities while simultaneously reducing false positives and operational costs. Notably, Argus has identified several critical zero-day vulnerabilities with CVE assignments.

## Recorded benchmark results

A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage.

- [Vulnerability Detection on PublicCMS](<https://www.sota2.com/research/sota/vulnerability-detection-on-publiccms.md>): Vulnerability Detection; PublicCMS; Detected Vulnerabilities Count: 6.

- [Vulnerability Detection on JeecgBoot](<https://www.sota2.com/research/sota/vulnerability-detection-on-jeecgboot.md>): Vulnerability Detection; JeecgBoot; Detected Vulnerabilities: 3.

- [Vulnerability Detection on Ruoyi](<https://www.sota2.com/research/sota/vulnerability-detection-on-ruoyi.md>): Vulnerability Detection; Ruoyi; Vuln Count: 3.

- [Vulnerability Detection on JSPWiki](<https://www.sota2.com/research/sota/vulnerability-detection-on-jspwiki.md>): Vulnerability Detection; JSPWiki; Detected Vulnerabilities Count: 5.

- [Vulnerability Detection on DataGear](<https://www.sota2.com/research/sota/vulnerability-detection-on-datagear.md>): Vulnerability Detection; DataGear; Vulnerabilities Detected: 3.

- [Vulnerability Detection on KeyCloak](<https://www.sota2.com/research/sota/vulnerability-detection-on-keycloak.md>): Vulnerability Detection; KeyCloak; Detected Vulnerabilities Count: 2.

- [Vulnerability Detection on Yudao-Cloud](<https://www.sota2.com/research/sota/vulnerability-detection-on-yudao-cloud.md>): Vulnerability Detection; Yudao-Cloud; Tokens: 85661.

## Original sources and implementation

[Original paper](<https://arxiv.org/abs/2604.06633>)

[PDF](<https://arxiv.org/pdf/2604.06633.pdf>)
