shunfeng8421/mcp-agent-security-papers
Runtime-verified MCP/agent security research: execution control-plane trust inversion, tool annotation trust labels, 4 confirmed disclosures
🎯 Vulnerability: Web Security Vulnerability
💡 Defense Fix: Implement strict input validation, contextual output encoding, and enforce least-privilege authorization controls.