Web deploymentWeb 版部署

Web deploymentWeb 版部署

Start with engine capabilities, then follow its installation, model sync and workspace guide. Pi remains experimental; engine capabilities differ.先了解各引擎能力，再按目标完成安装、模型同步和工作台配置。Pi 仍为实验版，各引擎功能并不相同。

Web 版部署指南

ClawPanel Web 版部署在 Linux 服务器上，通过浏览器管理 OpenClaw、Hermes、DSH、OpenCode 和实验版 Pi。各引擎能力不同，安装与生命周期独立。

Web 版的「本地」是服务端：模型测试、CLI 安装、工作目录和文件工具在运行 ClawPanel 的服务器或容器执行，不在浏览器电脑执行。填写 127.0.0.1 指向服务端回环地址；Docker 项目目录须单独挂载。共享登录和工作目录不是操作系统沙箱或多用户执行隔离。

部署方式一：官方一键脚本

以下脚本来自 ClawPanel 官方仓库。完整引擎环境推荐 Node.js 24.16.0+（24.x）和 npm；新安装默认选择官方稳定版 OpenClaw，汉化版可显式选择。部署面板不代表已安装全部引擎，完成后到各引擎页面继续设置。

curl -fsSL https://raw.githubusercontent.com/qingchencloud/clawpanel/main/scripts/linux-deploy.sh | bash

部署完成后访问 http://服务器IP:1420。完整说明见 Linux 部署指南。

部署方式二：手动部署

  1. 安装 Node.js 24.16.0+（24.x）
# Ubuntu/Debian
curl -fsSL https://deb.nodesource.com/setup_24.x | sudo -E bash -
sudo apt-get install -y nodejs

# 验证
node --version
  1. 按需安装 OpenClaw 官方稳定基线（其他引擎在各自页面安装）
npm install -g openclaw@2026.9.8 --registry https://registry.npmmirror.com
openclaw --version
  1. 克隆并安装 ClawPanel
cd /opt
sudo git clone https://github.com/qingchencloud/clawpanel.git
sudo chown -R $(whoami) clawpanel
cd clawpanel
npm install
npm run build
  1. 启动服务
cd /opt/clawpanel
npm run serve -- --port 1420
  1. 配置 systemd（推荐）
[Unit]
Description=ClawPanel Web
After=network.target

[Service]
Type=simple
WorkingDirectory=/opt/clawpanel
ExecStart=/usr/bin/node scripts/serve.js --port 1420
Environment=PATH=/usr/local/bin:/usr/bin:/bin
Restart=always
RestartSec=5

[Install]
WantedBy=multi-user.target

完整 Web 包升级

使用 Release 的 web-x.y.z.zip 完整 Web 服务端包，整体更新前端、Node 后端及 WebSocket 代理后重启，或重新构建镜像。frontend-hot-update-x.y.z.zip 仅供桌面前端热更新，不用于 Web 服务端部署；不要只覆盖 dist/。升级后检查 /__api/health 的前后端版本，并验证实际引擎连接。

安全建议

  • 使用 Nginx 反向代理 + HTTPS
  • 设置访问密码（在 ClawPanel 设置页面中配置）
  • 限制访问 IP 范围
  • 不要将端口直接暴露到公网

Web deployment

ClawPanel Web runs on a Linux server to manage OpenClaw, Hermes, DSH, OpenCode and experimental Pi from a browser. Engine capabilities, installations and lifecycles remain independent.

Local means server-side in Web mode: model tests, CLI installation, projects and file tools run on the panel server or container, not on your browser computer. 127.0.0.1 refers to the server loopback address. Mount Docker project directories separately; shared login and project paths are not an operating-system sandbox or multi-user execution isolation.

Official one-line script

The script comes from the official repository. Node.js 24.16.0+ (24.x) and npm are recommended for full engine use. New OpenClaw installs default to the official stable channel; the Chinese distribution is an explicit choice. Deploying the panel does not install every engine: continue in each engine's setup page.

curl -fsSL https://raw.githubusercontent.com/qingchencloud/clawpanel/main/scripts/linux-deploy.sh | bash

After deployment, open http://server-ip:1420. Read the full Linux deployment guide.

Manual deployment

# Install Node.js 24.x on Ubuntu/Debian; verify it is >=24.16.0
curl -fsSL https://deb.nodesource.com/setup_24.x | sudo -E bash -
sudo apt-get install -y nodejs

# Optional OpenClaw official baseline; install other engines in their own pages
npm install -g openclaw@2026.9.8 --registry https://registry.npmmirror.com
openclaw --version

# Clone and build
cd /opt
sudo git clone https://github.com/qingchencloud/clawpanel.git
sudo chown -R $(whoami) clawpanel
cd clawpanel
npm install
npm run build
npm run serve -- --port 1420

Full Web service upgrades

Use the complete web-x.y.z.zip service package, replace the frontend, Node backend and WebSocket proxy together, then restart; alternatively rebuild the image. frontend-hot-update-x.y.z.zip is a desktop frontend hot-update package, not a Web server deployment. Replacing only dist/ leaves mixed versions. Check frontend/backend versions at /__api/health and verify the engine connection after upgrading.

Production notes

  • Put the service behind Nginx, HTTPS and access control
  • Do not expose the raw port directly to the public Internet
  • Use systemd or another supervisor to keep the service running
qt.cool