Windows Automation

Automate Windows desktop by simulating mouse/keyboard input, managing clipboard, capturing screenshots, running commands, and launching apps.

Security Scan

The skill appears to implement the advertised Windows automation features, but a few inconsistencies and safety concerns (disabled pyautogui failsafe, no OS restriction despite Windows-specific behavior, and an API that runs arbitrary PowerShell commands) warrant caution before installing.

Purpose & Capability

Name, description, SKILL.md and the __init__.py implementation align: the skill uses pyautogui/pywin32/pyperclip to control mouse/keyboard, clipboard, screenshots, app launch and run commands. Minor inconsistency: registry metadata lists no OS restriction, but the code and docs are Windows-specific (uses PowerShell, 'start' via cmd). The lack of an explicit Windows-only restriction is an oversight.

Instruction Scope

SKILL.md and __init__.py instruct/implement arbitrary command execution via windows_command (subprocess.run calling 'powershell -Command <user command>') which is expected for a 'run commands' feature but is high-risk because it can execute any shell command. The SKILL.md and code also disable pyautogui.FAILSAFE (noted in docs and set in code), which removes a common user abort mechanism and increases the risk of runaway automation; this is a safety concern even if coherent with the skill's goal.

Install Mechanism

There is no install spec (instruction-only), and the SKILL.md suggests installing pyautogui, pywin32, pyperclip via pip. This is a low-risk, common approach. Note: the package requirements are only mentioned in docs, not declared in registry metadata.

Credentials

The skill does not request environment variables, credentials, or config paths. The capabilities implemented (mouse/keyboard, clipboard, screenshots, launching apps, running commands) do not require secrets, so the absence of requested credentials is proportionate.

Persistence & Privilege

The skill does not request always:true and does not modify other skills or system-wide configs. Model invocation is enabled (platform default) which is expected. No elevated persistence privileges are requested.

What to consider before installing

This skill implements legitimate Windows automation functions, but take these precautions before enabling it: 1) Only install on a Windows machine — the skill uses PowerShell and Windows 'start' and may misbehave on other OSes. 2) Be cautious about using windows_command: it executes arbitrary PowerShell commands, so avoid passing untrusted input. 3) Note that pyautogui.FAILSAFE is disabled in the code; you will not have the usual mouse-corner abort. If you plan to run complex automation, consider modifying the code to keep FAILSAFE enabled or add a safe abort mechanism. Finally, verify the package source for pyautogui/pywin32/pyperclip and review any code changes before use.

Windows Automation Skill

Automate Windows desktop interactions using PyAutoGUI

Overview

This skill provides comprehensive Windows desktop automation capabilities. Use it when you need to:

• Simulate mouse clicks and keyboard input
• Automate GUI interactions in non-web applications
• Capture screenshots
• Manage clipboard content
• Launch applications
• Run system commands

Setup

The required packages are already installed:

• pyautogui - Core automation engine
• pywin32 - Windows API access
• pyperclip - Clipboard operations

If needed, install manually:

pip install pyautogui pywin32 pyperclip

---

Tools

🖱️ Mouse Control

windows_click

Click at specified screen coordinates.

windows_click(x=100, y=200, button="left", clicks=1)

Parameters:

Param	Type	Default	Description
x	float	required	X coordinate
y	float	required	Y coordinate
button	str	"left"	"left", "right", "middle"
clicks	int	1	Number of clicks

---

windows_double_click

Double click at position.

windows_double_click(x=500, y=300)

---

windows_right_click

Right click at position.

windows_right_click(x=100, y=200)

---

windows_move

Move mouse to position.

windows_move(x=800, y=600, duration=0.5)

Parameters:

Param	Type	Default	Description
x	float	required	X coordinate
y	float	required	Y coordinate
duration	float	0.2	Movement duration in seconds

---

⌨️ Keyboard Control

windows_type

Type text using keyboard.

windows_type(text="Hello World!", interval=0.05)

Parameters:

Param	Type	Default	Description
text	str	required	Text to type
interval	float	0.05	Delay between keystrokes

---

windows_press

Press a single key.

windows_press(key="enter")

Supported keys:

• Function keys: f1 - f24
• Special keys: enter, esc, tab, space, backspace, delete
• Arrow keys: up, down, left, right
• Modifiers: ctrl, alt, shift, win

---

windows_hotkey

Press keyboard shortcut combination.

windows_hotkey("ctrl", "c")   # Copy
windows_hotkey("ctrl", "v")   # Paste
windows_hotkey("alt", "f4")   # Close window
windows_hotkey("win", "r")    # Open Run dialog

---

📸 Screen Operations

windows_screenshot

Take a screenshot.

windows_screenshot(filename="screenshot.png")

Parameters:

Param	Type	Default	Description
filename	str	"screenshot.png"	Filename to save

---

ℹ️ Information

windows_cursor_position

Get current cursor position.

windows_cursor_position()
# Returns: {"x": 500, "y": 300}

---

windows_screen_size

Get screen resolution.

windows_screen_size()
# Returns: {"width": 1920, "height": 1080}

---

windows_system_info

Get basic system information.

windows_system_info()
# Returns: {
#   "platform": "Windows",
#   "version": "10.0.22631",
#   "screen_size": {"width": 1920, "height": 1080},
#   "cursor_position": {"x": 500, "y": 300}
# }

---

🪟 Applications

windows_launch_app

Launch a Windows application.

windows_launch_app(name="notepad")
windows_launch_app(name="calc")
windows_launch_app(name="excel")
windows_launch_app(name="C:\\Program Files\\App\\app.exe")

---

💻 System

windows_command

Run a PowerShell or cmd command.

windows_command(command="ipconfig")
windows_command(command="Get-Process | Select-Object -First 5")
windows_command(command="dir C:\\")

---

📋 Clipboard

windows_clipboard_read

Read clipboard content.

windows_clipboard_read()
# Returns: "clipboard content"

---

windows_clipboard_write

Write text to clipboard.

windows_clipboard_write(text="Hello from OpenClaw!")

---

Usage Examples

Example 1: Open Notepad and Type Text

User: Open notepad and type hello
→ windows_launch_app with name="notepad"
→ windows_type with text="Hello World!"

Example 2: Fill a Form

User: Click on the name field and enter John
→ windows_click x=300 y=200
→ windows_type text="John"
→ windows_press key="tab"
→ windows_type text="[e-mail]"

Example 3: Screenshot and Save

User: Take a screenshot
→ windows_screenshot filename="screen_20260101.png"

Example 4: Copy File Path

User: Copy C:\Documents\file.txt to clipboard
→ windows_clipboard_write text="C:\\Documents\\file.txt"

Example 5: Get System Info

User: What's my screen resolution?
→ windows_screen_size

---

Safety Notes

• ⚠️ The mouse failsafe is disabled for smoother operation
• 📍 Screen coordinates start at (0, 0) in top-left corner
• 🖥️ Multi-monitor setups may have different coordinate systems
• 🔒 Some applications may block simulated input (games, secure apps)

---

Dependencies

Package	Purpose
pyautogui	Core automation
pywin32	Windows API
pyperclip	Clipboard

---

Built for OpenClaw 🦞