AI Security Agent: Multi-MCP Code Vulnerability Scanner
First benchmark testing LLM guards on safety and accuracy.