Netskope adds security for AI agent protocol MCP

Netskope has unveiled new security capabilities for its One platform specifically designed to manage the emerging risks associated with agentic artificial intelligence that utilizes the Model Context Protocol, SiliconANGLE reports. The update provides organizations with critical visibility and control as AI agents transition from pilot projects to production environments, where they can directly connect to enterprise data and workflows.

A core feature is the automatic discovery and identification of both MCP servers and clients on corporate networks, providing detailed metadata on each component. The platform also introduces granular policy controls, allowing security teams to define which AI agents can access specific resources and block unauthorized activity by default.

To assess risk, Netskope applies its Cloud Confidence Index to MCP-powered tools, while integrated data loss prevention monitors for the transfer of sensitive information like source code or credentials. According to Chief Product Officer John Martin, "MCP creates new security risks that legacy tools can’t solve," necessitating these new real-time inspection and auditing features.