Skip to main content
Glama

JCAppleScript

A Swift package for executing AppleScript from macOS applications, featuring a built-in MCP server that lets AI assistants control macOS apps through pre-built command shortcuts.

Overview

JCAppleScript provides three components:

  1. JCAppleScript (library) - Core AppleScript execution engine

  2. AppShortcuts (library) - Registry of pre-built AppleScript commands for popular macOS apps

  3. jcas-mcp (executable) - MCP (Model Context Protocol) server for AI-driven app automation

Related MCP server: osascript MCP Server

Installation

Swift Package Manager

Add JCAppleScript to your Package.swift:

dependencies: [
    .package(url: "https://github.com/johnnyclem/JCAppleScript.git", from: "2.0.0")
]

Then add the targets you need:

.target(
    name: "YourTarget",
    dependencies: [
        "JCAppleScript",     // Core engine only
        "AppShortcuts",      // App command registry
    ]
)

Quick Start

Using the Core Engine

import JCAppleScript

let engine = AppleScriptEngine.shared

// Execute raw AppleScript
let result = try engine.execute("""
    tell application "Finder"
        display dialog "Hello from Swift!"
    end tell
""")

// Send a command to an application
let output = try engine.tell(application: "Music", command: "play")

// Execute a script file with variable substitution
let fileResult = try engine.executeFile(at: "/path/to/script.scpt", variables: ["Alice", "Hello!"])

// Execute JavaScript for Automation (JXA)
let jxa = try engine.execute("Application('Music').play()", language: .javaScript)

// Check syntax without executing
try engine.checkSyntax("tell application \"Finder\" to activate")

// All execution APIs also have async variants
let asyncResult = try await engine.execute("return 40 + 2")

When embedding untrusted values in script source, escape them first:

let userInput = "…"
let script = "display dialog \(AppleScriptString.quoted(userInput))"

Using App Shortcuts

import AppShortcuts

let registry = AppRegistry.shared

// Execute a pre-built command
let result = try registry.executeCommand("messages.send_message", arguments: [
    "recipient": "+15551234567",
    "message": "Hello from JCAppleScript!"
])

// Discover available commands
let commands = registry.commands(forApp: "Reminders")
for cmd in commands {
    print("\(cmd.id): \(cmd.name) - \(cmd.description)")
}

// Search across all apps
let results = registry.searchCommands("send")

Using the MCP Server

The jcas-mcp executable is a Model Context Protocol server that AI assistants (Claude, GPT, etc.) can use to control macOS applications.

It is published to the official MCP registry as io.github.johnnyclem/jcas-mcp, and each GitHub release ships a prebuilt jcas-mcp.mcpb bundle (universal macOS binary) that can be installed directly in Claude Desktop via Settings → Extensions.

Setup with Claude Desktop

Add to your Claude Desktop config (~/Library/Application Support/Claude/claude_desktop_config.json):

{
  "mcpServers": {
    "applescript": {
      "command": "/path/to/jcas-mcp"
    }
  }
}

Build the server:

swift build -c release
# Binary at: .build/release/jcas-mcp

Available MCP Tools

Tool

Description

execute_applescript

Execute arbitrary AppleScript code †

execute_jxa

Execute JavaScript for Automation (JXA) code †

tell_application

Send a command to a specific app via tell block †

check_script_syntax

Compile a script (AppleScript or JXA) without executing it

list_running_applications

Get currently running applications

list_registered_apps

Browse all registered app command sheets

search_commands

Search for commands by keyword

run_app_command

Execute a pre-built command by ID

preview_app_command

Dry-run: show the exact script a command would execute

get_app_commands

Get detailed command info for a specific app

† Hidden and disabled when the server runs in safe mode (see below).

Server Configuration

Environment variable

Effect

JCAS_SAFE_MODE=1

Disables the arbitrary-code tools (execute_applescript, execute_jxa, tell_application) and registry commands flagged dangerous (e.g. terminal.run_command, safari.run_javascript). Only pre-built, sanitized registry commands remain available.

JCAS_APP_MANIFESTS=a.json:b.json

Colon-separated JSON manifest files with additional community app definitions to load at startup.

CLI flags: jcas-mcp --manifest prints the full registry as JSON, --version prints the server version, --help shows usage.

Example AI Interaction

User: "Send a message to John saying I'll be late"
AI uses tool: run_app_command
  command_id: "messages.send_message"
  arguments: { "recipient": "John", "message": "I'll be late" }

Supported Applications

JCAppleScript ships with command sheets for 12 built-in macOS apps:

App

Category

Commands

Examples

Messages

Communication

6

Send message, list chats, get participants

Mail

Communication

7

Compose email, search, check mail, list accounts

Reminders

Productivity

7

Create/complete/delete reminders, search, list

Calendar

Productivity

6

Create events, list today's events, upcoming

Notes

Productivity

8

Create/search/append notes, manage folders

Finder

System

12

File operations, folder contents, labels, trash

Safari

Internet

10

Open URLs, manage tabs, run JavaScript, get page content

Music

Media

13

Playback control, playlists, library search, ratings

Terminal

Development

8

Run commands, manage windows/tabs, profiles

System Settings

System

13

Dark mode, volume, notifications, dialogs, system info

Xcode

Development

20+

Open/build/run/test projects, schemes, build logs, debugging

Speech Recognition

System

3

Listen for spoken phrases via the system speech engine

Adding Custom App Support

Implement the ScriptableApp protocol to add support for any scriptable macOS app:

import AppShortcuts

struct MyApp: ScriptableApp {
    static let bundleIdentifier = "com.example.myapp"
    static let appName = "MyApp"
    static let description = "My custom application"
    static let category = AppCategory.productivity

    static let commands: [AppCommand] = [
        AppCommand(
            id: "myapp.do_thing",
            name: "Do Thing",
            description: "Performs the thing",
            parameters: [
                CommandParameter(name: "input", description: "The input value"),
            ]
        ) { args in
            let input = args["input", default: ""]
            return """
            tell application "MyApp"
                do thing with "\(input)"
            end tell
            """
        },
    ]
}

// Register at runtime
AppRegistry.shared.register(MyApp.self)

Community App Registry

JCAppleScript is designed to grow through community contributions. The app shortcut system uses a standard protocol (ScriptableApp) that makes it easy to:

  • Add new applications - Implement ScriptableApp for any scriptable macOS app

  • Extend existing apps - Submit new commands for already-registered apps

  • Share command sheets - Export/import app definitions via JSON manifests

We're building a browsable registry (similar to npmjs.org) where you can:

  • Browse applications and their supported AppleScript commands

  • Submit new commands for existing apps

  • Add entirely new applications to the registry

  • Generate JSON manifests for integration with other tools

Exporting and Importing the Registry

// Export all registered apps as manifest JSON
let json = try AppRegistry.shared.exportManifestJSON()

// Import community app definitions from a JSON manifest.
// Manifest commands are declarative script templates with ${param}
// placeholders; argument values are sanitized before substitution.
try AppRegistry.shared.loadManifest(contentsOf: URL(fileURLWithPath: "community.json"))

Example manifest:

[
  {
    "name": "CoolApp",
    "bundleIdentifier": "com.example.coolapp",
    "description": "A community-contributed app",
    "category": "Productivity",
    "commands": [
      {
        "id": "coolapp.greet",
        "name": "Greet",
        "description": "Show a greeting",
        "script": "tell application \"CoolApp\"\n    greet \"${who}\"\nend tell",
        "parameters": [
          {"name": "who", "description": "Who to greet", "required": true, "type": "string"}
        ]
      }
    ]
  }
]

The MCP server loads extra manifests from the JCAS_APP_MANIFESTS environment variable at startup.

Security Model

Registry command arguments are sanitized before script generation:

  • String, file-path, and date arguments are escaped (\, ", and control characters) so they cannot break out of AppleScript string literals.

  • Integer and boolean arguments are strictly validated/normalized — malformed values are rejected at validation and fall back to declared defaults during generation.

  • Values outside a parameter's allowedValues list are dropped.

  • Arguments that don't correspond to a declared parameter are discarded.

Commands that execute caller-supplied code (Terminal shell commands, Safari JavaScript) are flagged dangerous and can be disabled wholesale with JCAS_SAFE_MODE=1. Use the preview_app_command tool to inspect the exact script a command will run before executing it.

Note that execute_applescript, execute_jxa, and tell_application execute arbitrary code by design — only expose them to clients you trust, or run the server in safe mode.

Architecture

JCAppleScript/
├── Sources/
│   ├── JCAppleScript/           # Core engine
│   │   ├── AppleScriptEngine.swift
│   │   ├── AppleScriptSanitizer.swift
│   │   ├── ScriptResult.swift
│   │   └── ScriptError.swift
│   ├── AppShortcuts/            # App command registry
│   │   ├── AppProtocol.swift    # ScriptableApp protocol
│   │   ├── AppCommand.swift     # Command & parameter types + sanitization
│   │   ├── AppDefinition.swift  # Instance-based app description
│   │   ├── AppManifest.swift    # JSON import/export
│   │   ├── AppRegistry.swift    # Central registry
│   │   └── Apps/                # Built-in app sheets
│   │       ├── MessagesApp.swift
│   │       ├── RemindersApp.swift
│   │       ├── FinderApp.swift
│   │       ├── SafariApp.swift
│   │       ├── MailApp.swift
│   │       ├── CalendarApp.swift
│   │       ├── NotesApp.swift
│   │       ├── MusicApp.swift
│   │       ├── TerminalApp.swift
│   │       └── SystemSettingsApp.swift
│   └── JCAppleScriptMCP/       # MCP server
│       ├── main.swift
│       ├── MCPServer.swift
│       ├── MCPTransport.swift
│       └── MCPTypes.swift
├── Tests/
├── Legacy/                      # Original Obj-C implementation
├── Package.swift
└── LICENSE

Requirements

  • macOS 13.0+

  • Swift 5.9+

Legacy

The original Objective-C implementation (2013) is preserved in the Legacy/ directory for reference. It provided basic NSAppleScript wrapping with variable substitution. The new Swift implementation builds on those concepts while adding the MCP server, app registry, and modern Swift patterns.

License

MIT License - Copyright (c) 2013 John Clem. See LICENSE for details.

Related MCP Connectors

Related MCP Servers

  • A
    license
    B
    quality
    D
    maintenance
    Enables AI agents to interact with macOS applications (Finder, Mail, Contacts, Reminders, Notes, Calendar, TextEdit) using AppleScript. Allows AI assistants to perform tasks like searching contacts, managing files, checking email, and creating reminders through natural language.
    24
    1
    MIT
  • F
    license
    A
    quality
    D
    maintenance
    Enables the execution of AppleScript and JavaScript for Automation (JXA) on macOS to control applications, system events, and shell commands. It provides native automation capabilities with a security layer that permits most operations while specifically blocking file deletion commands.
    1
    6
    -
  • A
    license
    A
    quality
    C
    maintenance
    Enables Large Language Models to automate macOS applications and execute AppleScript commands through natural language. It features built-in security protections including application allowlists and dangerous pattern detection to prevent unauthorized or risky system operations.
    2
    MIT
  • A
    license
    Not graded
    quality
    F
    maintenance
    Enables AI assistants to automate macOS desktop tasks including mouse control, keyboard input, screenshots, window management, and UI interaction.
    6 npm
    415
    MIT