steipete/macos-automator-mcp

What it solves

It allows AI agents to control and automate a macOS system. By providing a bridge between an LLM and macOS's native scripting languages, it enables agents to interact with applications, inspect system states, and perform UI tasks that would otherwise be inaccessible to a standard chat interface.

How it works

The project implements a Model Context Protocol (MCP) server that exposes two primary tools to an AI client:

• get_scripting_tips: Allows the agent to search a built-in or local knowledge base of AppleScript and JavaScript for Automation (JXA) snippets to find the right way to perform a task.
• execute_script: Runs the actual script (either inline, from a file, or from the knowledge base) using the privileges of the host process.

Who it’s for

Developers and users who want to give their AI agents the ability to operate their Mac, automate repetitive workflows across different macOS apps, or perform system-level inspections.

Highlights

• Built-in Knowledge Base: Includes hundreds of pre-made AppleScript and JXA tips for browsers, terminals, and productivity apps.
• Customizable: Supports a local knowledge base to add or override bundled scripts.
• Flexible Execution: Supports both AppleScript and JXA.
• MCP Compliant: Easily integrates with any MCP-compatible client via npx.