AgentLure
Emerging1papers using it
2026first seen
AgentLure is a benchmark that contains context-dependent tasks and context-aware prompt injection attacks, used to evaluate the effectiveness of defenses against these security threats in Large Language Model (LLM) agents.