Show HN: Agentsec – Security scanner for AI agent installations (MCP, OpenClaw) (github.com)

Agentsec, a newly introduced security scanner for AI agent installations, targets common vulnerabilities and misconfigurations associated with agentic AI environments. It specifically evaluates systems like OpenClaw and MCP server configurations, focusing on risks such as credential exposure and plugin integrity. This tool is significant for the AI/ML community as it maps its findings to the OWASP Top 10 for Agentic Applications, promoting robust security practices amid increasing AI adoption. Key features of Agentsec include a comprehensive scanning capability that can pinpoint vulnerabilities ranging from gateway exposure to credential leak detection. Users can execute specific scans and generate reports in various formats, such as JSON and SARIF, suitable for CI/CD pipeline integration. The ability to harden configurations and set customized profiles for different deployment scenarios (like public bots or workstations) underscores Agentsec’s versatile application, making it a critical resource in enhancing the security posture of AI-driven platforms.