Fetching the paper…

Mimicking the Familiar: Dynamic Command Generation for Information Theft Attacks in LLM Tool-Learning System · Around