AgentSeal/awesome-mcp-security

# Technical Summary Employs a multi-layered static and dynamic analysis architecture with 9 specialized scanners—including Claude-powered semantic LLM classification, pattern-based detection for command injection and SSRF, and active "Deep Autopsy" probing with 45+ MCP-specific attack vectors—to quantify agent-accessible attack surface independent of code quality. Integrates across the MCP ecosystem to provide threat intelligence on 800+ Model Context Protocol servers, organizing findings by functional category with daily automated re-analysis and granular risk stratification from Safe to Dangerous.