mac-node-snapshot_skill

Overview

This skill provides a robust, permission-friendly method to capture macOS screens by recording a short video and extracting a high-quality PNG frame. It uses node screen.record to create a 1-second clip, then ffmpeg to pull a clean still image. The approach avoids many direct screencapture permission pitfalls and works well in headless or remote contexts. It returns a reliable image file ready to attach to replies.

How this skill works

The skill commands a node to record a short video of the display using OpenClaw's screen.record with no audio. It stores the output mp4 in a temporary folder, then runs ffmpeg to extract a single frame as a PNG. If the resulting frame is black, the workflow prompts the user to wake or unlock the screen and retries. The skill uses read on the generated PNG to attach the image to the response.

When to use it

• User asks for a screenshot or wants to know what is currently on their screen
• Direct screencapture tools fail due to macOS permission dialogs
• You need a reliable image capture in headless or remote sessions
• You want an automated way to capture and attach a screen image

Best practices

• Confirm ffmpeg is available before running; ask the user before installing
• Create and reuse a temporary folder (e.g., {skill}/tmp) for outputs
• If the capture is black, instruct the user to wake or unlock the display and retry
• Check node connectivity with nodes status when screen.record fails
• Respect privacy and ask for explicit user consent before capturing screens

Example use cases

• User requests a screenshot but native screencapture is blocked by macOS
• Automated diagnostics: capture screen state while troubleshooting an app
• Headless device management where GUI access is limited
• Archiving a UI state as part of a bug report or backup process