Add AI-Infra-Guard to Code section

[boy-hack]

AI Agent ecosystems are growing explosively, but security is lagging behind. AI-Infra-Guard by Tencent Zhuque Lab takes an attacker's perspective to detect security risks in AI products.

Coverage across 4 dimensions:

• Agent Security Scanning: Detects tool poisoning, data exfiltration, indirect prompt injection (9 risk classes)
• MCP/Skill Security Scanning: Static code analysis + AI agent-driven detection (command injection, SSRF, path traversal)
• AI Infrastructure CVE Detection: 50+ AI components (Ollama, ComfyUI, vLLM, LiteLLM), 1,200+ CVE fingerprints
• LLM Jailbreak Evaluation: Red-teaming LLMs for jailbreak resistance

Notable recognition:

• Black Hat Europe 2025 Arsenal + official talk "MCP Unchained"
• Recommended in DeepSeek's official README Security section
• Cited in 19 arXiv security papers
• 3,600+ Stars, MIT license

https://github.com/Tencent/AI-Infra-Guard