AI/LLM security scanner — model artifact analysis, prompt injection firewall, MCP agent validation, pickle/safetensors/GGUF fuzzing. Zero false positives.接入配置该项目没有提供统一的启动命令,请按源码仓库的说明安装。