Screenshot

Screenshot skill captures pixel-perfect screenshots of web pages and passes them to Claude's vision capabilities for analysis. Capture full-page screenshots, visible viewport only, or specific DOM elements targeted by CSS selector. Output formats include PNG, JPEG, WebP, and PDF for multi-page sites.

Visual monitoring workflows: check if a website's UI changed since yesterday, verify that a deployed feature renders correctly across different viewport sizes (mobile/tablet/desktop), or detect visual regressions in a web application after an update. The skill supports scheduled captures, saving screenshots with timestamps to a local directory for historical comparison.

When paired with Claude's vision, the captured screenshots become queryable: 'Is the checkout button still green?', 'Did the price change on this product page?', 'Is the site showing an error page?'. This enables visual intelligence that pure HTML scraping cannot provide — catching CSS bugs, layout breaks, or content changes that don't appear in the page source.

The skill also supports capturing authenticated pages when used in conjunction with Browser Relay sessions, and handles cookie consent banners automatically by accepting them before capture to ensure clean screenshots. Pixel density control lets you capture Retina-quality 2x screenshots for high-fidelity visual documentation. QA teams, UX researchers, and site reliability engineers use Screenshot for automated visual testing pipelines integrated with their CI/CD workflows.

Installation

clawhub install screenshot

Install: clawhub install screenshot