Overview

A.I.G (AI-Infra-Guard) is a comprehensive, full-stack AI Red Teaming platform designed by Tencent Zhuque Lab to rigorously secure AI ecosystems. It empowers organizations with advanced tools for self-examination of AI security risks, encompassing infrastructure, agents, and Large Language Models (LLMs). By integrating multiple specialized scanning and evaluation modules, A.I.G offers a proactive approach to identifying and mitigating vulnerabilities before they can be exploited.

Key Features

• Agent Scan: An independent, multi-agent automated scanning framework to evaluate the security of AI agent workflows across platforms like Dify and Coze.
• Skills Scan (aig-skill-scan): Leverages advanced LLMs to audit AI agent skills against 9 categories of security risks defined by SkillTrustBench, achieving high F1 scores with leading models. It offers granular risk detection from instruction hijacking to insecure coding practices.
• MCP Server & Agent Skills Scan: Detects 14 major security risk categories in both MCP Servers and Agent Skills, supporting scans from source code and remote URLs.
• AI Infra Vulnerability Scan: Precisely identifies over 100 AI framework components and covers more than 2000 known CVE vulnerabilities in popular tools like Ollama, ComfyUI, vLLM, and n8n.
• LLM Jailbreak Evaluation: Assesses prompt security risks using curated datasets and multiple attack methods to test LLM robustness, providing detailed cross-model comparison.
• ClawScan (OpenClaw Security Scan): Enables one-click evaluation of OpenClaw security risks, including insecure configurations, skill risks, CVE vulnerabilities, and privacy leakage.
• Model and API Relay Checker: Features model fingerprinting, Claude Signature verification, relay black-box auditing, and more.
• Extensible Plugin Framework: Facilitates community contributions through fingerprint rules, vulnerability rules, MCP plugins, and jailbreak evaluation datasets.
• Modern Web Interface & Comprehensive API: Offers a user-friendly UI for one-click scanning and real-time progress, alongside a complete API with Swagger specifications for seamless integration.

Typical Use Cases

• Proactive AI Infrastructure Security Auditing: Continuously scan deployed AI infrastructure (e.g., vLLM, Ollama, ComfyUI) for known CVE vulnerabilities and misconfigurations to prevent breaches.
• AI Agent & Skill Security Assurance: Integrate A.I.G into CI/CD pipelines to automatically assess the security posture of custom AI agent skills and MCP servers, ensuring compliance and mitigating risks before deployment.
• LLM Prompt Security Testing: Evaluate the resilience of LLMs against adversarial attacks, such as jailbreaking, to understand and defend against prompt injection and manipulation techniques.
• Security Verification for AI Development Teams: Provide developers and security professionals with a powerful, all-in-one platform to red team their AI applications and systems, identifying weaknesses across the entire AI lifecycle.
• Open-Source AI Security Community Engagement: Leverage A.I.G's open-source nature to contribute to and benefit from community-driven security enhancements for AI technologies.