computer-use-bridge

Controls macOS GUI apps programmatically via command line with dual-backend automation support.

5|Updated Mar 27, 2026
One-click install
npx skills add https://github.com/barkleesanders/claude-code-starter --skill computer-use-bridge
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: computer-use-bridge
Source: https://github.com/barkleesanders/claude-code-starter/tree/main/skills/computer-use-bridge
Command: npx skills add https://github.com/barkleesanders/claude-code-starter --skill computer-use-bridge

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Manually clicking through macOS apps to complete repetitive desktop tasks is slow and inefficient. This Skill lets you control any native Mac application programmatically from the command line, no custom per-app automation scripts required.

Core Features & Use Cases

  • Dual backend support: Choose between OpenAI Codex Computer Use for vision-based, foreground (including locked screen) automation, and cua-driver for background, cursor-free accessibility-tree automation.
  • Built-in utility commands: Run health checks to verify TCC permissions and backend setup, list all running apps instantly, and get machine-readable JSON output for agent workflows.
  • Common use cases: Automate native app workflows like organizing files in Finder or filling out productivity app forms, take over GUI tasks when you can't interact with your Mac manually, and run background automation without interrupting your current work.

Quick Start

Use the computer-use-bridge skill to open your Documents folder in Finder, create a new folder named Invoices, and move all PDF files downloaded in the last week into that new folder.

Frequently Asked Questions about computer-use-bridge

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I automate macOS GUI applications from the command line?

You can automate macOS GUI applications from the command line by using this Skill to simulate inputs, launch apps, and execute cross-app workflows programmatically without manual clicking. It eliminates repetitive desktop interactions by supporting both vision-based and accessibility-tree automation.

Can I control Mac desktop apps in the background without moving my cursor?

Yes, you can control Mac desktop apps in the background without moving your cursor by using the cua-driver backend. This approach uses accessibility-tree automation to execute tasks without interrupting your current foreground work or taking over your screen.

What's the best way to automate native Mac app workflows like organizing Finder files?

The best way to automate native Mac app workflows like organizing Finder files is to use this Skill's utility commands to programmatically open directories, create folders, and move files based on specific criteria. It provides machine-readable JSON output to integrate these actions into larger agent workflows.

Do I need special permissions to run GUI automation scripts on macOS?

Yes, GUI automation on macOS requires valid TCC permissions for both screen recording and accessibility access. You can run a built-in health check command to verify that your system has the correct permissions and backend setups configured properly.

How does vision-based foreground automation differ from accessibility-tree automation for Mac?

Vision-based foreground automation uses OpenAI Codex Computer Use to interact with the screen visually, even supporting locked screens. Accessibility-tree automation uses cua-driver to interact with application elements in the background without requiring visual focus or cursor movement.