Show HN: AgentTrust – A–F trust scores for MCP servers (50 graded, open source) (github.com)

AgentTrust has recently launched an open-source platform designed to enhance the trust and security of autonomous AI agents and MCP (Model Context Protocol) servers by providing verifiable Trust Cards and grade badges. This tool allows users to scan their AI agent codebases for vulnerabilities, aligning findings with the OWASP Agentic AI Top 10 and LLM Top 10 for 2025. The platform runs entirely locally, ensuring zero data retention and no telemetry, while providing essential outputs such as machine-readable Trust Cards, SARIF reports for GitHub integration, and shareable markdown badges. This initiative is significant for the AI/ML community as it addresses pressing security concerns amid growing reliance on AI technologies by offering a systematic approach to vulnerability detection across key areas such as shell injection and secret leaks. AgentTrust’s capabilities to evaluate workflow reliability and integrate with CI/CD pipelines provide a robust solution for organizations aiming to ensure compliance with upcoming regulations, like the EU AI Act. Moreover, the ease of embedding trust scores into README files offers a transparent way for developers to showcase the security credentials of their AI systems, fostering trust among stakeholders.