N/A

elliotllliu/AgentShield

mcp agent Offline

Full-stack security for AI agents — static analysis + MCP runtime interception. 31 rules detect prompt injection, data exfiltration, backdoors, tool poisoning, and cross-file attack chains. Includes MCP proxy for real-time blocking, Python AST taint tracking, multi-language injection detection (8 languages), and AI-powered deep analysis. Free, offline, zero-config.

Scan Scheduled

This agent is queued for security scanning. It will be graded in the next scan batch.

What We Know

Browse more: Search all agents Ecosystem Report