As reported by Silicon Angle, Virtue AI Inc. has introduced AgentSuite, a new software platform designed to protect artificial intelligence agents from a range of cyber threats, including malicious MCP servers. This launch follows the company's significant initial funding round less than a year ago.
AgentSuite addresses vulnerabilities where AI agents interact with external applications via MCP servers. The platform scrutinizes tool definitions within MCP servers for malicious prompts, such as attempts to extract training data. It also scans MCP server code for known vulnerabilities. A key component, the Unified Agent Gateway, acts as an intermediary, enforcing company cybersecurity policies on tool use requests with minimal latency. AgentSuite further enhances security by simulating cyberattacks in over 50 sandboxed environments to identify weaknesses in both the agents and their associated MCP servers. Integration with CI/CD systems allows for the flagging of vulnerabilities before AI agent updates are deployed.
By simulating attacks and integrating with development pipelines, AgentSuite aims to prevent vulnerabilities from reaching production, addressing the complex security challenges posed by agentic AI.
Source: Silicon Angle
