pip install -r requirements.txt
推荐使用uv环境
uv init MCP  
cd MCP
echo "3.13" > .python-version
uv venv
source .venv/bin/activate
uv add "mcp[cli]" httpx

python main.py
使用uv安装：uv run main.py
服务器默认运行在 `http://0.0.0.0:8000/sse`

通过 MCP 客户端（如 Cherry Studio/Kiro/Cursor/Trae）连接：
{
  "mcpServers": {
    "Security-Scanner": {
      "url": "http://10.0.xx.xx:8000/sse"
    }
  }
}

对该MCP进行安全扫描
http://xxx.xxx.xxx:7777/sse

对该MCP进行安全扫描
https://xxx.xxx.xxx/mcp-servers/plant-monitor-getproductionorder/sse
Authorization=Bearer apikey-693xxxxxxxxxxxxx

├── agent.py          # ReAct 智能体实现
├── config.py         # 攻击载荷和检测规则配置
├── graph.py          # LangGraph 工作流定义
├── llm.py            # LLM 客户端（支持通义千问/Azure OpenAI）
├── server.py         # FastMCP 服务器
├── state.py          # 状态定义
├── task_manager.py   # 异步任务管理器
├── tools.py          # 安全检测工具函数
└── system_prompt.md  # 系统提示词

{
  "status": "completed",
  "target": "http://example.com/sse",
  "risk_level": "HIGH",
  "summary": {
    "tools_discovered": 10,
    "injectable_tools": 5,
    "attacks_executed": 30,
    "vulnerabilities_found": 3
  },
  "vulnerabilities_by_severity": {
    "HIGH": {
      "count": 2,
      "details": [...]
    },
    "MEDIUM": {
      "count": 1,
      "details": [...]
    }
  }
}