3 papers
cs.CR2026
ToolRosetta: Surface-Form Invariance of Tool-Call Guards
Hasan Demirkiran, Jens Ernstberger
Step-level guards for tool-using agents are evaluated on tidy JSON tool calls, but deployed coding agents act through execute_bash and execute_code tools whose payloads are strings…
cs.CR2026
Language Is a Security Surface for Agent Guards
Hasan Demirkiran, Jens Ernstberger
Step-level guards decide, before each tool call, whether an agent's next action is safe given the user's request and the trajectory so far. Users write those requests in many langu…
cs.AI2026
Shortcut Sensitivity in Agent-Safety Evaluations
Hasan Demirkiran, Jens Ernstberger
AI agents take actions through tools. Some actions are sensitive, so guards that supervise them must themselves be evaluated reliably. Yet the data used to evaluate a guard can inc…