altic-studio - SKILL.md Agent Skill

name: altic-studio description: macOS automation skill for AppleScript actions and Chrome browser control via MCP CDP tools. license: Apache-2.0

Altic Studio

altic-studio provides two automation modes:

AppleScript mode for macOS apps and system actions
MCP CDP mode for Google Chrome browser control
MCP file mode for safe Finder and filesystem operations
MCP clipboard mode for text, file, and image pasteboard operations
MCP window/workspace mode for arranging macOS apps and windows

It also includes Swift utility scripts for active-display screenshots, clipboard file/image operations, and window/workspace management on macOS.

Mode A: AppleScript (macOS apps)

Always execute AppleScript through Bash with osascript.
Always run from workspace root as the working directory.
Quote script paths and arguments.
Prefer direct script invocation; do not route through Python wrappers unless explicitly requested.

Command templates:

osascript "skills/altic-studio/scripts/<script>.applescript" [arg1] [arg2] ...

AppleScript Capabilities

The full Altic automation surface is exposed as scripts under skills/altic-studio/scripts:

open-application.applescript - args: <app_name>
send-message.applescript - args: <phone_or_handle> <message>
read-recent-messages.applescript - args: <phone_or_handle> <count>
fetch-all-contacts.applescript - args: none
set-reminder.applescript - args: <text> <YYYY-MM-DD HH:MM> [list_name]
create-note.applescript - args: <title> <body> [folder]
search-for-note.applescript - args: <query> [max_results]
create-calendar-event.applescript - args: <title> <YYYY-MM-DD HH:MM> <duration_minutes> [calendar]
list-all-calendar-events-for-day.applescript - args: <YYYY-MM-DD>
open-safari-tab.applescript - args: [url]
close-safari-tab.applescript - args: [tab_index|-1]
get-safari-tabs.applescript - args: none
switch-safari-tab.applescript - args: <tab_index>
run-safari-javascript.applescript - args: <javascript_code>
navigate-safari.applescript - args: <url>
reload-safari-page.applescript - args: none
safari-go-back.applescript - args: none
safari-go-forward.applescript - args: none
open-safari-window.applescript - args: [url]
close-safari-window.applescript - args: none
get-safari-page-info.applescript - args: none
decrease-brightness.applescript - args: [amount_0_to_1]
increase-brightness.applescript - args: [amount_0_to_1]
turn-up-volume.applescript - args: [amount_0_to_100]
turn-down-volume.applescript - args: [amount_0_to_100]
capture-screenshot.applescript - args: [output_path] [full|interactive|window]
capture-active-screen.swift - args: <output_path> (captures full display containing frontmost app)
clipboard.swift - subcommands: get-files, set-files <paths...>, save-image <output_path>, set-image <image_path>
window-manager.swift - subcommands: get_frontmost_app, list_windows, focus_window, move_window, resize_window, center_window, tile_windows, minimize, hide_app, quit_app

Swift command template (for active-display screenshots):

swift "skills/altic-studio/scripts/capture-active-screen.swift" "/tmp/active-screen.png"

Swift command template (for window management):

swift "skills/altic-studio/scripts/window-manager.swift" "list_windows" '{"include_minimized":false}'

Mode B: Chrome Browser Control (MCP CDP)

Use MCP tools for deterministic Chrome automation:

chrome_open_session
chrome_navigate
chrome_wait_for
chrome_click
chrome_type
chrome_extract
chrome_screenshot
chrome_close_session
chrome_list_sessions
capture_active_screen

Execution pattern:

Open a Chrome session.
Navigate and wait for stable selectors.
Interact with click and type actions.
Verify state with extraction.
Capture screenshots on checkpoints or failures.
Close session.

Mode C: File Finder and File Operations (MCP)

Use MCP file tools instead of shell commands when the user asks to find, inspect, copy, move, rename, reveal, or trash files. These tools return JSON for successful lookups and operations, and Error: ... strings for failures.

Available tools:

find_files - args: <query> [root] [max_results] [include_hidden] [kind]
list_directory - args: <path> [include_hidden] [max_results]
get_file_info - args: <path>
copy_file - args: <source> <destination> [overwrite] [dry_run]
copy_directory - args: <source> <destination> [overwrite] [dry_run]
move_file - args: <source> <destination> [overwrite] [dry_run]
rename_file - args: <path> <new_name> [overwrite] [dry_run]
trash_file - args: <path> [dry_run]
reveal_in_finder - args: <path>
get_finder_selection - args: none

File workflow rules:

If the user refers to selected files, the current Finder window, or "this file", call get_finder_selection first.
For ambiguous file names, call find_files, then get_file_info; ask for disambiguation when multiple plausible matches remain.
Prefer dry_run=true before copy_file, copy_directory, move_file, rename_file, or trash_file when the target or destination is ambiguous.
Do not overwrite unless the user explicitly asks for replacement; default overwrite=false.
Use trash_file, not permanent deletion.
Use reveal_in_finder after file operations when the user wants to see the result in Finder.

Mode D: Clipboard Operations (MCP)

Use MCP clipboard tools instead of shell commands when the user asks to inspect, copy, paste, clear, or save clipboard contents. These tools return JSON for text and file operations, Error: ... strings for failures, and image content for saved clipboard images.

Available tools:

get_clipboard_text - args: [max_chars]
set_clipboard_text - args: <text>
clear_clipboard - args: none
get_clipboard_files - args: none
set_clipboard_files - args: <paths>
save_clipboard_image - args: [output_path]
set_clipboard_image - args: <path>

Clipboard workflow rules:

Ask before overwriting clipboard contents unless the user explicitly asks to copy, set, clear, or replace the clipboard.
Use get_clipboard_text first when the user asks what text is currently copied.
Use get_clipboard_files when the user says they copied files in Finder or wants to paste copied files somewhere else.
Use save_clipboard_image when the user wants to inspect, store, or transform an image currently on the clipboard.
Use set_clipboard_files with existing paths only; resolve ambiguous file names with find_files before changing the clipboard.
Use set_clipboard_image only for existing image files.

Mode E: Window and Workspace Operations (MCP)

Use MCP window tools when the user asks to arrange the workspace, focus an app or window, tile apps side by side, resize windows, minimize windows, hide apps, quit apps, or inspect the frontmost app. These tools return JSON for successful operations and Error: ... strings for failures.

Available tools:

get_frontmost_app - args: none
list_windows - args: [app_name] [include_minimized]
focus_window - args: [app_name] [window_id] [window_index]
move_window - args: <x> <y> [app_name] [window_id] [window_index] [display_index]
resize_window - args: <width> <height> [app_name] [window_id] [window_index] [display_index]
center_window - args: [app_name] [window_id] [window_index] [display_index] [width] [height]
tile_windows - args: [layout=columns|rows|grid] [app_names] [display_index] [padding]
minimize - args: [app_name] [window_id] [window_index]
hide_app - args: <app_name>
quit_app - args: <app_name>

Window workflow rules:

Call list_windows before moving, resizing, minimizing, or focusing when the target app has multiple windows or the target is ambiguous.
Prefer app_name for user-facing workflows and window_id for precise follow-up operations after list_windows.
Use display_index when arranging windows on a specific display; otherwise tools choose the display containing the target window.
Use tile_windows with explicit app_names when preparing a multi-app task.
Do not call quit_app unless the user explicitly asked to close or quit that app.
If a window mutation fails with an Accessibility error, tell the user to grant Accessibility permission to the host app running the MCP server.

Operational Rules

Validate date/time format before running reminder/calendar scripts.
If contact lookup returns multiple options, ask for disambiguation before sending.
If script output indicates permission issues, report exact permissions to enable.
Prefer explicit CSS selectors and call chrome_wait_for before click/type on dynamic pages.
After mutating actions in Chrome, verify expected page state with chrome_extract.
For file mutations, verify the target with get_file_info or list_directory after the operation when the user needs confirmation.
For clipboard mutations, verify with get_clipboard_text, get_clipboard_files, or save_clipboard_image when the user needs confirmation.
For window mutations, verify with list_windows when the user needs confirmation.

Permissions Checklist

Contacts access
Calendars access
Reminders access
Automation permission for app control
Accessibility permission for system controls and window management
Screen Recording permission for screenshots and improved window discovery
Safari setting: Allow JavaScript from Apple Events
Google Chrome installed for CDP tools
Full Disk Access for reading Messages database
Automation permission for Finder when revealing, trashing, or reading selection