| name | screenshot |
| description | This skill should be used when the user asks to "take a screenshot", "capture a page", "screenshot a website", or invokes /screenshot. Takes a URL as argument. |
| version | 0.2.0 |
Take a website screenshot
Capture a clean, high-resolution (1920x1080) screenshot of a web page.
Setup (once)
cd .claude/skills/screenshot/scripts && npm install
Usage
Run the script via Bash:
node .claude/skills/screenshot/scripts/screenshot.js <url> [output.png] [--full]
url (required): the page URL
output.png (optional): output file path, defaults to {hostname}.png
--full (optional): capture the full scrollable page instead of just the viewport
The script automatically removes cookie banners and chat widgets before taking the screenshot (see scripts/cleanup.js).
How it works
- Launches headless Chromium at 1920x1080
- Navigates to the URL (waits for network idle)
- Runs DOM cleanup (
scripts/cleanup.js): removes cookie banners (Cookiebot, OneTrust, GDPR, etc.), chat widgets (Crisp, Intercom, Drift, HubSpot, Chatwoot, etc.), and small fixed-position corner elements
- Waits 500ms for layout reflow
- Takes the screenshot and prints the output path
Extending cleanup
To add new selectors for cookie banners or chat widgets, edit scripts/cleanup.js.