Tencent/AI-Infra-Guard

A full-stack AI Red Teaming platform securing AI ecosystems via OpenClaw Security Scan, Agent Scan, Skills Scan, MCP scan, AI Infra scan and LLM jailbreak evaluation.

What it solves

AI-Infra-Guard (A.I.G) is an AI red teaming platform designed to help enterprises and individuals identify security risks in their AI infrastructure, agents, and models. It provides a centralized solution for self-examining vulnerabilities in AI frameworks, agent workflows, and prompt robustness.

How it works

The platform integrates several specialized scanning and evaluation tools:

  • AI Infrastructure Scan: Fingerprints live AI services (like vLLM, Ollama, and ComfyUI) to detect known CVE vulnerabilities across over 100 components.
  • MCP Server & Agent Skills Scan: Analyzes source code or remote URLs to detect 14 categories of security risks in Model Context Protocol (MCP) servers and agent skills.
  • Agent Scan: An automated multi-agent framework that evaluates the security of agent workflows on platforms like Dify and Coze.
  • Jailbreak Evaluation: Tests the robustness of LLMs by applying curated attack datasets and multiple attack methods to identify prompt security risks.
  • ClawScan: A one-click evaluation tool for OpenClaw security risks, including insecure configurations and privacy leaks.

Who it’s for

It is intended for security researchers, AI developers, and enterprises deploying AI agents and infrastructure who need to perform security audits and red teaming exercises.

Highlights

  • Comprehensive Coverage: Detects 1900+ known CVEs and covers 68+ AI components.
  • Extensible Architecture: Uses a plugin framework allowing users to add their own fingerprint, vulnerability, and jailbreak rules.
  • CI/CD Integration: Includes aig-skill-scan for integrating agent skill security audits into development pipelines.
  • User-Friendly Interface: Provides a modern web UI for one-click scanning and real-time progress tracking.