Guard

mcp-guard prevents critical security vulnerabilities in AI agent interactions by comprehensively scanning Model Context Protocol server configurations for common flaws like leaked secrets, injection vulnerabilities, and misconfigurations before they are deployed. It also provides a robust runtime gateway to enforce security policies on every call to an MCP server, acting as a crucial defense layer against potential attacker footholds that could arise from compromised tool descriptions or sensitive data exposure.

주요 기능

사용 사례