A bridge for controlling Windows GUI environments from Linux-based AI agents using screen capture and peripheral simulation.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install pc-control
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install pc-control using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Prefer to download?
Get the raw skill files in a ZIP archive.
What is PC Control?
PC Control is a specialized automation bridge designed to empower AI agents with the ability to interact directly with a Windows desktop environment from WSL or Linux. By leveraging a FastAPI server running on the Windows host and a Python-based client, it enables seamless remote control through mouse and keyboard simulation. This tool is a vital part of the Openclaw Skills ecosystem, allowing developers to automate tasks in graphical applications that lack command-line interfaces.
The skill operates by capturing high-resolution screenshots for visual analysis, which are then used by the agent to determine the necessary cursor movements or keystrokes. This approach ensures that legacy software, complex design tools, and Windows-specific settings are fully accessible to your automated workflows.
PC Control Use Cases
- Automating interactions with proprietary GUI software that lacks an API or CLI.
- Remote troubleshooting and system configuration of Windows hosts from a Linux terminal.
- Performing visual regression testing or UI audits across desktop applications.
- Streamlining multi-app workflows that require switching between browser and desktop tools.
- Providing visual context to LLMs for autonomous desktop navigation and task completion.
How PC Control Works
- Launch the FastAPI server on the Windows host system using the integrated launcher script.
- Establish a secure connection from the WSL environment to the Windows server via a local network socket.
- Capture the current screen state using the screenshot utility to provide visual feedback to the AI agent.
- Analyze the screenshot to identify target elements and calculate precise X/Y coordinates.
- Dispatch mouse clicks, drags, scrolls, or keyboard input commands to the Windows server.
- Execute a verification screenshot after each action to confirm the desired state change occurred.
PC Control Setup
First, configure the config.json file in the skill directory to point to your Windows Python executable. Then, install the required dependencies using the provided script:
python3 scripts/install.py
To begin using the Openclaw Skills interface for desktop control, start the server:
python3 scripts/launcher.py start
PC Control Data Schema & Taxonomy
The skill manages configuration and session-specific data to ensure secure and efficient execution:
| Component | Format | Purpose |
|---|---|---|
| config.json | JSON | Defines server ports, Python paths, and default screenshot scaling factors. |
| Screenshots | JPG/PNG | Temporary visual data used for coordinate mapping and visual reasoning. |
| Token Auth | String | Automatically generated session tokens to secure the FastAPI endpoint. |
| Scale Factor | Float | Multiplier used to translate image coordinates back to physical screen pixels. |
PC Control Advanced Features
- Intelligent screenshot scaling to optimize performance for various vision models.
- Support for complex hotkey combinations and system-level shortcuts like Win+R.
- Automated server lifecycle management with configurable auto-shutdown timers.
- Token-based authentication ensuring only authorized local clients can control the desktop.
- Direct integration with Openclaw Skills workflows for multi-agent GUI interaction.
SKILL.md
Loading
Related Openclaw Skills
ComfyUI Painter
A professional-grade bridge for local ComfyUI image generation and automated CivitAI model lifecycle management within the Openclaw Skills ecosystem.
Aviation Agent
A dedicated flight briefing assistant that fetches real-time aviation weather and provides FAA regulatory guidance.
System Monitor
A specialized monitoring utility for tracking local server hardware performance including CPU, RAM, and GPU metrics.
ZeroDust Chain Exit
A powerful utility to sweep 100% of native gas tokens from EIP-7702 compatible chains, leaving exactly zero balance.
Task Engine
A sophisticated multi-agent orchestration engine that uses a state machine to track, dispatch, and monitor complex project tasks across various AI agents.
Personalized News Collector
An intelligent agent skill that analyzes user history to deliver a prioritized, interest-based news report.
Featured*
