Mac Compute Use for Openclaw

A comprehensive skill for controlling macOS applications and automating GUI interactions through the Accessibility API.

reg2005
v0.1.0
Mar 2, 2026
1
1.5k
0

Install & Download

1. ClawHub CLI

The fastest way to install a skill directly from the registry.

npx clawhub@latest install mac-compute-use

2. Manual Installation

Copy the skill folder to one of these locations

Global
~/.openclaw/skills/
Workspace
<project>/skills/

Priority: Workspace > Local > Bundled

3. Prompt Installation

Copy this prompt to OpenClaw to install it automatically.

Help me install mac-compute-use 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 Mac Compute Use?

Mac Compute Use is a powerful extension for Openclaw Skills that enables AI agents to interact directly with the macOS desktop environment. By utilizing the system's Accessibility API, this skill provides a way to programmatically navigate, inspect, and control any graphical application running on Mac. Whether it is managing files in Finder, sending messages, or interacting with specialized software, this tool bridges the gap between digital reasoning and physical UI interaction. For users of Openclaw Skills, this provides a computer use capability that allows for the automation of complex, manual desktop tasks that do not offer a traditional CLI or API interface.

Mac Compute Use Use Cases

  • Automating repetitive workflows in native macOS applications like Mail, Messages, and Finder.
  • Performing web automation and testing in browsers without requiring specialized extensions.
  • Extracting data from legacy or proprietary software that only provides a graphical interface.
  • Coordinating multi-app workflows, such as moving data from a web browser into a desktop spreadsheet.
  • Enabling Accessibility-driven UI testing and verification for developers.

How Mac Compute Use Works

  1. The skill initiates a session by opening or activating a target macOS application using its name or bundle identifier.
  2. It captures the current UI state of the application, identifying all interactive elements and their precise screen coordinates.
  3. The agent processes this UI tree to determine the necessary interaction, such as a click, keystroke, or scroll action.
  4. The skill executes the requested interaction at the exact coordinates provided by the Accessibility API.
  5. After each action, the skill returns a diff of the UI state, allowing Openclaw Skills to verify changes and proceed with the next step in the automation.

Mac Compute Use Setup

First, install the MCP server via Homebrew:

brew tap reedburns/mcp-server-macos-use
brew install mcp-server-macos-use

Next, register the server with mcporter to enable it within your Openclaw Skills environment:

mcporter config add macos-use --transport stdio --command $(which mcp-server-macos-use)

Finally, you must manually grant Accessibility permissions by navigating to System Settings → Privacy & Security → Accessibility and adding the mcp-server-macos-use binary.

Mac Compute Use Data Schema & Taxonomy

The skill manages its data through structured JSON responses and temporary files that define the UI tree.

Property Description
status The outcome of the tool execution (success or error).
pid The Process ID of the application currently being controlled.
file The absolute path to a JSON file containing the full UI element tree.
visible_elements A summary of interactive elements, including their coordinates (x, y) and dimensions.
diff A report of what changed in the UI following the last interaction.

Mac Compute Use Advanced Features

  • Support for complex keyboard modifiers including Command, Option, Shift, and Control.
  • Coordinate-based clicking with optional width and height parameters to target element centers accurately.
  • Bi-directional scrolling capabilities for navigating long documents or web pages.
  • Application traversal that identifies bundle IDs, process IDs, and element roles for precise targeting.
  • Integration with local search tools like grep or Python to parse deep UI hierarchies within Openclaw Skills.

SKILL.md


Loading

Related Openclaw Skills

METADATA

Github Stars: 0
forks: 0

Featured*