plugin-vetting
MCP server for security-vetting third-party AI agent extensions before installation — Claude skills, plugins, tool packs. 41 detection rules across prompt-injection, exfiltration, dynamic execution, typosquats. 0-100 risk score.
