Windows下载OpenClaw源码，启动和安装攻略

OpenClaw交流
, , ,
1

安装向导

  • 1.安装node，推荐版本24，最低版本22.16+，推荐使用nvm控制node版本
  • 2.安装pnpm npm install -g pnpm
  • 3.克隆并构建
    • git clone https://github.com/openclaw/openclaw.git
    • cd openclaw
    • pnpm install
    • pnpm ui:build
    • pnpm build
  • 4.链接cli，让openclaw命令全局可用 pnpm link --global

验证：命令行输入openclaw，有相关提示

报错及解决方法

  • 用管理员打开命令行/powershell

  • pnpm install报错Error error code 3221225477

    • 降级node到22版本
    • 安装C++构建工具
      • 下载 Visual Studio Build Tools 2022
    • 运行安装器，工作负载列表中只勾选“使用 C++ 的桌面开发”
    • 完成后需重启电脑，重新install即可（建议删掉node_modules和pnpm-lock.yaml并执行pnpm store prune）

  • pnpm build报错scripts/bundle-a2ui.sh: line 31: node: command not found

    • 创建scripts/bundle-a2ui.mjs
// scripts/bundle-a2ui.mjs
// OpenClaw A2UI Bundle Placeholder Generator
// For public repository users who do not have access to private A2UI source code.
// This script creates a minimal valid ES module to satisfy TypeScript compilation.

import fs from "node:fs";
import path from "node:path";
import { createHash } from "node:crypto";
import { fileURLToPath } from "node:url";

// ── Resolve project root directory correctly on Windows and Unix ──
const __filename = fileURLToPath(import.meta.url);
const __dirname = path.dirname(__filename);
const ROOT_DIR = path.resolve(__dirname, ".."); // openclaw/ root

// ── Define output paths ──
const OUTPUT_DIR = path.join(ROOT_DIR, "src", "canvas-host", "a2ui");
const OUTPUT_FILE = path.join(OUTPUT_DIR, "a2ui.bundle.js");
const HASH_FILE = path.join(OUTPUT_DIR, ".bundle.hash");

// ── Ensure output directory exists ──
fs.mkdirSync(OUTPUT_DIR, { recursive: true });

// ── Generate placeholder content (valid ES module) ──
const placeholderContent =
  `

// Auto-generated placeholder for A2UI

// Source code is not available in the public OpenClaw repository.

// This file exists only to satisfy build dependencies.

export const A2UI = {

version: '0.0.0-placeholder',

render: () => {

throw new Error('A2UI runtime is not available in this build.');

}

};

`.trim() + "\n";

// ── Write the bundle file ──
fs.writeFileSync(OUTPUT_FILE, placeholderContent);

// ── Compute and write hash to prevent unnecessary rebuilds ──
const hash = createHash("sha256").update(placeholderContent).digest("hex");
fs.writeFileSync(HASH_FILE, hash);

// ── Success message ──
console.log("✅ A2UI placeholder bundle created successfully.");
console.log(` Bundle: ${OUTPUT_FILE}`);
console.log(` Hash: ${HASH_FILE}`);
  • 在package.json中替换脚本
{
  "scripts": {
    //...
    "canvas:a2ui:bundle": "node --import tsx scripts/bundle-a2ui.mjs"
  }
}

新手向导

openclaw onboard --install-daemon

  • I understand this is personal-by-default and shared/multi-user use requires lock-down. Continue?（风险告知）
    • Yes
  • Setup mode（选择配置模式）
    • QuickStart
  • Model/auth provider（选择执行模型）
    • Skip for now（暂时跳过）
  • Filter models by provider（选择供应商）
    • All providers
  • Default model （默认模型）
    • Keep current

验证：全部显示:white_check_mark:即表示配置成功。

启动向导

  • 1.运行诊断命令 pnpm openclaw doctor
    • 依次提示“是否创建token”、“是否创建.openclaw文件夹”等，均选择Yes
    • 完成后，会在C:\Users\${userName}\里创建.openclaw文件夹
  • 2.配置本地网关 openclaw config set gateway.mode local
  • 3.启动本地网关 openclaw gateway
  • 4.在浏览器打开（默认http://127.0.0.1:18789）

验证：可在浏览器打开初始化连接界面

相关配置

  • 配置文件位置：C:\Users\${userName}\openclaw.json
  • 如已一键安装过OpenClaw，建议备份整个.openclaw文件夹，并选择性覆盖

token

  • 初始化界面输入token并连接，然后在控制-概览里配置网关令牌

模型

// openclaw.json
{
  // ...,
  "models": {
    "providers": {
      "DeepSeek": {
        "baseUrl": "xxx", // 替换即可
        "apiKey": "xxx", // 替换即可
        "api": "openai-completions",
        "models": [
          {
            "id": "xx", // 替换即可
            "name": "DeepSeek-V3.2 (非思考模式)",
            "reasoning": false,
            "input": ["text"],
            "cost": {
              "input": 0,
              "output": 0,
              "cacheRead": 0,
              "cacheWrite": 0
            },
            "contextWindow": 128000,
            "maxTokens": 8192
          }
        ]
      }
    }
  }
}
2

减少api调用次数是关键

3

比训练专用模型便宜多了