VirusTotal is expanding its Crowdsourced AI program by integrating AgentMesh—a specialized technology from Knostic designed to secure the agentic layer.
What Happened
As part of its Crowdsourced AI program, VirusTotal is integrating AgentMesh technology, which focuses on analyzing the security of tools interacting with AI agents. The system enables the detection of supply chain threats, including prompt injections and malicious behavior in MCP servers, VS Code extensions, and other tools used by autonomous assistants.
Context
With the rise of autonomous AI assistants, the cybersecurity focus is shifting from traditional malicious code analysis to protecting interaction tools. This creates a need to analyze tool definitions and their behavior, as classical signature-based analysis methods are not always effective against threats hidden within the logic of the agentic layer.
Why It Matters for the Industry
This integration marks the formation of a new cybersecurity vertical market for Agentic Workflows. It transforms the security auditing of MCP servers and AI tools from a niche task into an industry standard, creating a specialized line of defense for tools that interact directly with LLMs.
Why It Matters for Users
Users of AI coding assistants and developers utilizing third-party extensions or MCP servers gain a higher level of trust in their tools. VirusTotal's expertise helps mitigate the risks of using malware via prompt injections and provides more reliable verification of third-party extensions.
Sources
Author
Look at AI, Editorial Team