6 papers · 1 filter
Securing Web Agents with the Compile-Then-Execute Paradigm
Annabella Chow, Jinhao Zhu, Sylvie Venuto +5
Web agents commonly follow the ReAct paradigm, selecting each action from runtime observations that inherently mix trusted and untrusted content. This design makes their control fl…
Pincer: Resource Authorization for Agents using a Digital Twin
Mayank Rathee, Alexander Stepanov, Shalin Madabhavi +3
Coding agents have become increasingly long-horizon, autonomous, reliant on general-purpose shell and maintain their own persistent memory for self-improvement. While these capabil…
Repeat-After-Me: Black-Box Adaptive Visual Prompt Injection
Sizhe Chen, Yu-Lin Tsai, Ivan Evtimov +4
Prompt injection is widely recognized as a major security threat to AI agents that interact with untrusted external data, such as websites, documents, and emails. Prior work has sh…
Daydreaming: Stealing Hidden Agent Skills through Black-Box Task Interaction
Yu-Lin Tsai, Yu-An Lu, Ci-Yang Tsai +3
Agent skills bundle instructions, reference data, and executable helpers that let a general agent perform specialized tasks. Hosted providers can keep these files secret while sell…
Web Agents Should Adopt the Compile-and-Execute Paradigm
Julien Piet, Annabella Chow, Yiwei Hou +5
ReAct has become the default architecture across LLM agents, and many existing web agents follow this paradigm. We argue that it is the wrong default for web agents. Instead, web a…
MiniScope: Authorizing Agents with Least-Privilege Permissions
Jinhao Zhu, Kevin Tseng, Gil Vernik +4
AI agents are increasingly granted autonomous access to sensitive user data and third-party services, making effective permission management a critical security challenge. Existing…