Protectt.ai has launched a Model Context Protocol (MCP) Security Solution aimed at addressing security risks associated with the use of agentic AI in banking, financial services and insurance (BFSI), fintech and other enterprises.
The solution follows a “Scan, Test, Protect” framework covering MCP servers and their interactions with AI agents and enterprise tools. Protectt.ai said the platform is designed to identify vulnerabilities before adoption, test systems before deployment and monitor interactions after deployment.
The company said its MCP Scanner examines MCP server packages, tool definitions and parameter schemas for risks including poisoned tool descriptions, hidden instructions, vulnerable dependencies and supply-chain threats. Its MCP Security Testing component conducts automated simulations to identify vulnerabilities involving tool squatting, injection and information leakage.
The MCP Proxy provides an inline security layer between AI agents and external tools, inspecting tool calls and responses in real time. It can detect injected instructions, block unauthorised tools, redact personally identifiable information and credentials, and generate audit logs for export to security information and event management (SIEM) platforms.
“The security conversation around AI has largely focused on the model. But as agentic AI becomes autonomous, it is potentially more exposed to threats when connected to tools and enterprise systems,” said Manish Mimani, Co-founder and CEO of Protectt.ai. “Organisations must take cognizance of these evolving vulnerabilities as threats evolve. Our full-stack MCP Security Solution plugs this gap by providing unified protection across the agent’s lifecycle.”