A Formal Security Framework for MCP-Based AI Agents: Threat Taxonomy, Verification Models, and Defense Mechanisms

Acharya, Nirajan, Gupta, Gaurav Kumar · arXiv · 2026