A zero-trust gateway for securely brokering interactions between AI models and internal tools via the Model Context Protocol, with DLP, prompt injection defense, and LLM-as-a-Judge.
