Install
openclaw skills install openclaw-computer-use-v2-0-0-tarEnable OpenClaw to control and use the computer like a human. Use when: (1) User asks to open applications or files, (2) User needs to automate desktop tasks...
openclaw skills install openclaw-computer-use-v2-0-0-tar🖥️ 让 OpenClaw 像人类一样使用电脑
"给你的 AI 一双手,让它真正操作你的电脑"
版本: 2.0.0 Pro | 更新: 2026-03-29 | 新增: 增强API、图片识别、进度显示、批量操作
| Situation | Action |
|---|---|
| 打开应用程序 | 使用 computer-open-app 命令 |
| 截图/录屏 | 使用 computer-screenshot 或 computer-record |
| 文件管理 | 使用 computer-file-manager |
| 鼠标键盘控制 | 使用 computer-mouse 或 computer-keyboard |
| 自动化任务 | 使用 computer-automation 脚本 |
| 系统监控 | 使用 computer-monitor |
┌─────────────────────────────────────────┐
│ OpenClaw Computer Use │
│ 🖥️ 电脑控制中枢 │
├─────────────────────────────────────────┤
│ │
│ 🖱️ GUI 自动化 │
│ - 鼠标控制(移动、点击、拖拽) │
│ - 键盘输入(文字、快捷键) │
│ - 窗口管理(打开、关闭、切换) │
│ │
│ 📁 文件管理 │
│ - 浏览文件夹 │
│ - 复制/移动/删除文件 │
│ - 搜索文件 │
│ - 批量重命名 │
│ │
│ 📸 屏幕捕获 │
│ - 全屏截图 │
│ - 区域截图 │
│ - 窗口截图 │
│ - 屏幕录制 │
│ │
│ 🚀 应用控制 │
│ - 启动应用程序 │
│ - 关闭应用程序 │
│ - 应用间切换 │
│ - 获取应用信息 │
│ │
│ ⚙️ 系统监控 │
│ - CPU/内存/磁盘监控 │
│ - 进程管理 │
│ - 网络监控 │
│ - 系统日志 │
│ │
└─────────────────────────────────────────┘
# Ubuntu/Debian
sudo apt-get update
sudo apt-get install -y \
scrot \
xdotool \
wmctrl \
xclip \
ffmpeg \
python3-tk \
python3-dev
# macOS
brew install \
imagemagick \
cliclick \
ffmpeg
# Python 依赖
pip install \
pyautogui \
pillow \
opencv-python \
pynput \
psutil
# 通过 skillhub
skillhub install openclaw-computer-use
# 或手动安装
git clone https://github.com/openclaw/openclaw-computer-use.git \
~/.openclaw/workspace/skills/openclaw-computer-use
# 全屏截图
computer-screenshot --full --output ~/screenshots/desktop.png
# 区域截图(交互式选择区域)
computer-screenshot --region --output ~/screenshots/region.png
# 特定窗口截图
computer-screenshot --window "Chrome" --output ~/screenshots/browser.png
# 连续截图(每隔5秒)
computer-screenshot --interval 5 --count 10 --output ~/timelapse/
Python API:
from computer_use import Screenshot
screenshot = Screenshot()
screenshot.full_screen(save_path="~/desktop.png")
screenshot.region(x=100, y=100, width=500, height=400)
screenshot.window(title="Terminal")
# 移动鼠标到指定位置
computer-mouse move --x 500 --y 300
# 点击
computer-mouse click --button left
computer-mouse double-click
computer-mouse right-click
# 拖拽
computer-mouse drag --start-x 100 --start-y 100 --end-x 400 --end-y 400
# 滚动
computer-mouse scroll --amount -5
# 键盘输入
computer-keyboard type "Hello, World!"
# 快捷键
computer-keyboard hotkey ctrl alt t # 打开终端
computer-keyboard hotkey alt tab # 切换窗口
computer-keyboard hotkey ctrl c # 复制
computer-keyboard hotkey ctrl v # 粘贴
Python API:
from computer_use import Mouse, Keyboard
mouse = Mouse()
mouse.move(500, 300)
mouse.click()
mouse.drag(100, 100, 400, 400)
keyboard = Keyboard()
keyboard.type("Hello, World!")
keyboard.hotkey('ctrl', 'c')
# 启动应用
computer-app launch --name "Google Chrome"
computer-app launch --name "code" --args "/path/to/project"
# 关闭应用
computer-app close --name "Chrome"
computer-app close --pid 12345
# 切换窗口
computer-app focus --name "Terminal"
computer-app list # 列出所有窗口
# 获取应用信息
computer-app info --name "Chrome"
Python API:
from computer_use import Application
app = Application()
app.launch("google-chrome")
app.focus("Chrome")
app.close("Chrome")
# 获取所有窗口
windows = app.list_windows()
for window in windows:
print(f"{window['title']} - {window['pid']}")
# 浏览目录
computer-file list ~/Documents
# 搜索文件
computer-file search --name "*.pdf" --path ~/Downloads
# 复制/移动/删除
computer-file copy ~/file.txt ~/backup/
computer-file move ~/old/ ~/new/
computer-file delete ~/temp/
# 批量重命名
computer-file rename --pattern "IMG_*.jpg" --format "Photo_{num:03d}.jpg"
# 获取文件信息
computer-file info ~/document.pdf
Python API:
from computer_use import FileManager
fm = FileManager()
files = fm.list_directory("~/Documents")
results = fm.search("*.pdf", "~/Downloads")
fm.copy("~/file.txt", "~/backup/")
# 录制全屏
computer-record --full --duration 60 --output ~/recording.mp4
# 录制特定区域
computer-record --region --duration 300 --output ~/demo.mp4
# 录制特定窗口
computer-record --window "VS Code" --output ~/coding.mp4
# 停止录制
computer-record --stop
# 查看系统资源
computer-monitor resources
# 查看进程
computer-monitor processes --top 10
# 杀死进程
computer-monitor kill --pid 12345
computer-monitor kill --name "chrome"
# 网络监控
computer-monitor network
# 磁盘使用
computer-monitor disk
Python API:
from computer_use import SystemMonitor
monitor = SystemMonitor()
cpu = monitor.cpu_percent()
memory = monitor.memory_info()
disk = monitor.disk_usage("/")
processes = monitor.top_processes(n=10)
#!/usr/bin/env python3
# scripts/open-workspace.py
from computer_use import Application, Mouse, Keyboard
import time
def open_workspace():
"""自动打开工作环境:浏览器+编辑器+终端"""
# 打开 Chrome
app = Application()
app.launch("google-chrome", args=["--new-window", "https://github.com"])
time.sleep(2)
# 打开 VS Code
app.launch("code", args=["~/projects/myapp"])
time.sleep(2)
# 打开终端
keyboard = Keyboard()
keyboard.hotkey('ctrl', 'alt', 't')
time.sleep(1)
keyboard.type("cd ~/projects/myapp && npm start")
keyboard.hotkey('return')
print("✓ 工作环境已就绪")
if __name__ == "__main__":
open_workspace()
#!/usr/bin/env python3
# scripts/capture-and-share.py
from computer_use import Screenshot
import datetime
def capture_screen():
"""截图并保存到指定目录"""
screenshot = Screenshot()
timestamp = datetime.datetime.now().strftime("%Y%m%d_%H%M%S")
filename = f"screenshot_{timestamp}.png"
# 全屏截图
path = screenshot.full_screen(save_path=f"~/Screenshots/{filename}")
print(f"✓ 截图已保存: {path}")
return path
if __name__ == "__main__":
capture_screen()
#!/usr/bin/env python3
# scripts/auto-test.py
from computer_use import Mouse, Keyboard, Screenshot
import time
def run_ui_test():
"""自动化 UI 测试"""
mouse = Mouse()
keyboard = Keyboard()
screenshot = Screenshot()
# 步骤1:打开应用
keyboard.hotkey('alt', 'f2')
time.sleep(0.5)
keyboard.type("myapp")
keyboard.hotkey('return')
time.sleep(3)
# 步骤2:截图记录初始状态
screenshot.window("MyApp", save_path="~/test/step1_initial.png")
# 步骤3:点击按钮
mouse.click(500, 400)
time.sleep(1)
# 步骤4:截图记录结果
screenshot.window("MyApp", save_path="~/test/step2_result.png")
print("✓ UI 测试完成")
if __name__ == "__main__":
run_ui_test()
权限控制
沙箱模式
操作日志
# ~/.openclaw/computer-use-config.yml
security:
# 需要确认的操作
require_confirmation:
- delete
- kill
- sudo
# 禁止访问的目录
forbidden_paths:
- /etc
- /usr/bin
- ~/.ssh
# 白名单应用
allowed_apps:
- google-chrome
- code
- terminal
- nautilus
screenshot:
default_save_path: ~/Screenshots
format: png
quality: 90
recording:
default_save_path: ~/Recordings
fps: 30
codec: h264
Q: 截图失败
# 检查依赖
which scrot # Linux
which import # ImageMagick
# 安装缺失依赖
sudo apt-get install scrot
Q: 鼠标控制无效
# 检查权限
echo $DISPLAY
# 应该输出 :0 或类似
# 授予权限
xhost +local:
Q: 应用无法启动
# 检查应用是否存在
which google-chrome
which code
# 使用完整路径
computer-app launch --path "/usr/bin/google-chrome"
# 在 OpenClaw 中使用
# 用户:"帮我打开工作区"
def open_workspace():
from computer_use import Application
app = Application()
app.launch("code")
app.launch("google-chrome")
return "✓ 工作区已打开"
# 用户:"截图给我看"
def take_screenshot():
from computer_use import Screenshot
screenshot = Screenshot()
path = screenshot.full_screen()
return f"截图已保存: {path}"
# 每小时截图记录工作状态
0 * * * * /path/to/computer-screenshot --full --output ~/timelapse/
# 每天自动清理旧截图
0 0 * * * find ~/Screenshots -mtime +7 -delete
| 特性 | 说明 |
|---|---|
| 🖥️ 完整控制 | 鼠标、键盘、应用、文件全覆盖 |
| 🤖 自动化 | 脚本化重复任务 |
| 📸 可视化 | 截图、录屏记录操作 |
| 🔒 安全 | 权限控制、操作日志 |
| 🐍 Python API | 易于集成和扩展 |
现在,让 OpenClaw 真正控制你的电脑! 🖥️🚀
Skill Version: 1.0.0 Compatible with: OpenClaw 2026.3.24+, Linux, macOS