传统手动安装Ubuntu Server的典型流程（部分步骤）

 1. 下载ISO镜像
wget https://releases.ubuntu.com/22.04/ubuntu-22.04-live-server-amd64.iso

 2. 制作启动U盘
dd if=ubuntu-22.04-live-server-amd64.iso of=/dev/sdb bs=4M status=progress

 3. 手动安装过程（交互式）
 - 选择语言、键盘布局
 - 配置网络（静态IP、DNS）
 - 设置分区方案
 - 创建用户和密码
 - 选择要安装的软件包

 4. 安装完成后手动配置
sudo apt update && sudo apt upgrade -y
sudo apt install -y nginx mysql-server redis-server
sudo systemctl enable nginx
 ... 还要配置防火墙、SSH、监控等

 模拟装机AI助手的简化工作流程

class AIDeploymentAgent:
    def __init__(self, llm):
        self.llm = llm   大语言模型作为"大脑"
        self.tools = {
            "execute_command": self.run_shell_command,
            "download_file": self.download,
            "edit_config": self.modify_config,
            "check_service": self.check_service_status
        }
    
    def execute(self, user_instruction: str):
         Step 1: LLM理解用户意图并生成执行计划
        plan = self.llm.generate_plan(user_instruction)
         计划示例: ["检查是否已安装Python3", "如未安装则执行安装", "创建虚拟环境", "安装requirements.txt"]
        
         Step 2: Agent逐步执行计划
        for step in plan:
            action = step["action"]       如 "execute_command"
            params = step["params"]       如 {"cmd": "python3 --version"}
            result = self.tools[action](params)
            
             Step 3: 根据执行结果决定下一步
            if result["status"] == "error":
                 LLM根据错误信息生成修复方案
                fix_plan = self.llm.generate_fix_plan(result["error"])
                 执行修复...
        
        return "部署完成"

!/usr/bin/env python3
 minimal_deploy_agent.py - 最小化装机AI助手示例

import subprocess
import json
from typing import Dict, List, Optional

class MinimalDeployAgent:
    """一个简化的装机AI助手"""
    
    def __init__(self):
         注册可用工具
        self.tools = {
            "run": self._run_command,       执行shell命令
            "check": self._check_exists,    检查软件是否已安装
            "install": self._install_pkg    安装软件包
        }
    
    def _run_command(self, cmd: str) -> Dict:
        """执行shell命令并返回结果"""
        try:
            result = subprocess.run(cmd, shell=True, capture_output=True, text=True, timeout=300)
            return {
                "status": "success" if result.returncode == 0 else "error",
                "stdout": result.stdout,
                "stderr": result.stderr
            }
        except Exception as e:
            return {"status": "error", "message": str(e)}
    
    def _check_exists(self, pkg_name: str, pkg_manager: str = "apt") -> bool:
        """检查软件包是否已安装"""
        cmd_map = {
            "apt": f"dpkg -l | grep -q {pkg_name}",
            "pip": f"pip3 show {pkg_name} > /dev/null 2>&1"
        }
        result = subprocess.run(cmd_map.get(pkg_manager, cmd_map["apt"]), 
                                shell=True, capture_output=True)
        return result.returncode == 0
    
    def _install_pkg(self, pkg_name: str, pkg_manager: str = "apt") -> Dict:
        """安装软件包"""
        cmd_map = {
            "apt": f"sudo apt install -y {pkg_name}",
            "pip": f"pip3 install {pkg_name}"
        }
        return self._run_command(cmd_map.get(pkg_manager, cmd_map["apt"]))
    
    def deploy(self, requirements: List[Dict]) -> Dict:
        """
        执行部署任务
        requirements示例: [
            {"action": "check", "params": {"pkg_name": "python3", "pkg_manager": "apt"}},
            {"action": "install", "params": {"pkg_name": "nginx", "pkg_manager": "apt"}}
        ]
        """
        results = []
        for task in requirements:
            action = task["action"]
            params = task["params"]
            
            if action not in self.tools:
                results.append({"task": task, "status": "skipped", "reason": f"unknown action: {action}"})
                continue
            
            result = self.tools[action](params)
            results.append({"task": task, "result": result})
            
             关键步骤失败则终止
            if action == "install" and result.get("status") == "error":
                return {"overall": "failed", "results": results, "failed_at": task}
        
        return {"overall": "success", "results": results}


 ========== 使用示例 ==========
if __name__ == "__main__":
    agent = MinimalDeployAgent()
    
     定义部署任务清单（真实场景中由LLM根据自然语言生成）
    deploy_plan = [
        {"action": "check", "params": {"pkg_name": "python3", "pkg_manager": "apt"}},
        {"action": "install", "params": {"pkg_name": "nginx", "pkg_manager": "apt"}},
        {"action": "check", "params": {"pkg_name": "nginx", "pkg_manager": "apt"}},
        {"action": "install", "params": {"pkg_name": "flask", "pkg_manager": "pip"}}
    ]
    
    print("🚀 开始部署...")
    result = agent.deploy(deploy_plan)
    print(json.dumps(result, indent=2, ensure_ascii=False))