Setup & Installation
Or with the ClawHub CLI, for registry-managed skill folders outside a full OpenClaw workspace:
Version History
- Added detailed setup instructions for API keys, permissions, and customizations required before using the AI agent. - Updated usage guidance for `airpoint ask`, including output expectations and screenshots for confirmation. - Documented the new `airpoint stop` command to cancel running tasks. - Clarified the role of Gemini API key in enhancing on-screen target detection. - Modified examples and explanations for improved clarity and onboarding.
What This Skill Does
Airpoint is a macOS AI agent that sees and controls your screen using natural language. You give it a plain-English instruction and it opens apps, clicks UI elements, types, scrolls, and manages windows autonomously. It uses the accessibility tree, screenshots, and a visual locator to perceive and act on whatever is displayed.
Unlike keyboard macros or scripted automation, Airpoint can adapt to whatever is currently on screen rather than requiring predefined UI coordinates or app-specific scripts.
When to use it
- Ask the agent to find and summarize the latest email in Mail
- Enable a macOS system setting without navigating menus manually
- Search the web in Safari by describing what you want
- Read an on-screen error message and report its contents
- Chain multi-step tasks across different apps in one command
Example Workflow
Here's how your AI assistant might use this skill in practice.
User asks: open Slack and read the last unread message in the #general channel
- 1Runs `airpoint ask "open Slack and read the last unread message in the #general channel"`
- 2Agent opens Slack via Spotlight if not already running
- 3Agent navigates to the #general channel using the accessibility tree
- 4Agent reads the unread message text from the screen
- 5Returns a text summary and saves a screenshot of the result
A text summary of the message content and a screenshot path showing the Slack window with the message visible
Requirements
Accounts, API keys, or tools you or your AI assistant may need to set up while using this skill.