Search papers, labs, and topics across Lattice.
This paper introduces AgentGuard, an attribute-based access control framework designed to mitigate security risks in tool-use LLM agents. AgentGuard employs a client-server architecture with lightweight client-side integration and server-side inspection mechanisms to address both single-tool and cross-tool vulnerabilities. Evaluation demonstrates AgentGuard's ability to enforce security policies and audit agent behavior with minimal code modification.
Prevent LLM agents from leaking data or causing financial loss with AgentGuard, a new access control framework that requires only ~10 lines of code to integrate.
LLM-based agents have recently attracted significant attention due to their ability to autonomously invoke relevant tools to accomplish complex tasks. However, recent studies have shown that these agents face severe security risks, which may lead to privacy leakage, financial loss, or even full system compromise. In this paper, we present AgentGuard, an attribute-based access control framework for tool-use LLM-based agents. AgentGuard adopts a client-server architecture. On the client side, AgentGuard provides lightweight integration for agents implemented in different programming languages and architectures. It requires only minor code modifications (e.g., around 10 lines) without changing the underlying agent execution logic. On the server side, AgentGuard provides three complementary inspection mechanisms to cover both single-tool and cross-tool security risks in agent execution. In addition, it offers a visualized front-end interface for security policy specification and runtime auditing. Currently, AgentGuard is publicly accessible at https://github.com/WhitzardAgent/AgentGuard.
