Log-To-Leak: Prompt Injection Attacks on Tool-Using LLM Agents via the Model Context Protocol ↗
arXiv · Artificial Intelligence ·
Category intelligence
News & signals
Product announcements and research worth investigating. Every item links to its original source; none is an independently measured result.
Cross-User Poisoning: User-Task Boundary Failures in Multi-User Collaborative Language Agents ↗
arXiv · Artificial Intelligence ·
AgentSentry: Mitigating Indirect Prompt Injection in LLM Agents via Temporal Causal Diagnostics and Context Purification ↗
arXiv · Artificial Intelligence ·