Agent Audit : A Security Analysis System for LLM Agent Applications

Haiyue Zhang, Yi Nian, Yue Zhao · 2026

Read Agent Audit : A Security Analysis System for LLM Agent Applications by Haiyue Zhang, Yi Nian, Yue Zhao (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2603-22853.txt.

Description

Zhang, Nian, and Zhao present Agent Audit, a security scanner for language model agent apps. Many agent failures come from the software stack, not the model. Risky tool code, exposed credentials, and over-privileged settings are common sources. The scanner combines dataflow analysis, credential detection, and configuration checks. Findings appear in terminal, JSON, and SARIF formats for CI pipelines. On a benchmark with 42 known flaws, it finds 40 with six false positives. Scans run in under a second, and the tool is open source.

Catalog record

CREATORHaiyue Zhang, Yi Nian, Yue Zhao
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE23007
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2603.22853

Browse the full library · About the Tower of Babel