crawl-screenshot

Use the crawl_screenshot tool from the Crawlbase MCP server to capture a screenshot of a webpage.

When to use

Use this skill when the user wants a visual snapshot of a webpage, needs to verify a page's appearance, or wants to share a visual preview of content.

Usage

Call the crawl_screenshot tool with the url parameter. Optionally pass:

• device: "desktop" (default), "mobile", or "tablet"
• mode: "fullpage" (default) or "viewport"
• width: max width in pixels (viewport mode only)
• height: max height in pixels (viewport mode only)
• page_wait: milliseconds to wait before taking the screenshot
• store: true to push the underlying HTML page to Crawlbase Cloud Storage. The screenshot itself is NOT stored — the response still includes an ephemeral screenshot_url, plus rid, stored_at, storage_url, and token_type: "js". Screenshots always land in the JS-token silo, so subsequent storage_get / storage_delete calls must pass use_js_token: true.

Example prompts

• "Take a screenshot of https://example.com"
• "Capture a mobile screenshot of https://example.com"
• "Take a viewport screenshot of https://example.com at 1280x800"
• "Take a screenshot of https://example.com and store the page in Crawlbase Cloud Storage"