Chromium Screenshots

About

Taking screenshots for Vision AI agents often leads to page state drift when DOM and pixels are captured separately, causing inconsistencies. Chromium Screenshots solves this by guaranteeing 'Zero-Drift' capture, extracting both the DOM coordinates and the pixel-perfect screenshot from the exact same render frame. It acts as a visual cortex for AI agents, providing a robust API and MCP server for injecting full authentication states, waiting for page elements, and delivering comprehensive visual and structural data.

Key Features

• Zero-Drift screenshot and DOM extraction from the same render frame
• Full authentication state injection (cookies, localStorage, sessionStorage)
• Built-in quality assessment for captured data (Good/Low/Poor scores)
• Native Model Context Protocol (MCP) server for AI agent integration
• Robust SPA support with wait conditions (selectors, network idle)
• 0 GitHub stars

Use Cases

• Automating web interactions for AI agents via an MCP server
• Enhancing Vision AI accuracy with ground-truth DOM coordinates
• Capturing authenticated dashboards and complex web applications