Model context protocol (MCP) gives IT teams a standardized way to connect large language models (LLMs) to tools and data sources when developing AI-based workflows. But security researchers warn that ...