Website

You can use this MCP server to capture fast, high-quality screenshots of webpages with automatic tiling for AI vision workflows. It is optimized for models like Claude Vision, delivering perfectly sized 1072x1072 tiles and offering options for full-page captures, screencasts, and flexible viewport configurations.

How to use

Install and run the MCP server in your development environment, then use the provided tools to capture screenshots or create screencasts for AI workflows. The two core tools are take_screenshot for single or tiled captures and take_screencast for time-based screencasts. You can save results to a directory to obtain file paths, or receive base64-encoded images for immediate in-context use.

How to install

Prerequisites you need before installation. Make sure you have Node.js 20.x or higher, and npm or npx installed. Chrome/Chromium is required and will be downloaded automatically by the capture tool.

claude mcp add screenshot-website-fast -s user -- npx -y @just-every/mcp-screenshot-website-fast

If you prefer to install via a code editor integration, you can also add the MCP server using the following alternatives.

code --add-mcp '{"name":"screenshot-website-fast","command":"npx","args":["-y","@just-every/mcp-screenshot-website-fast"]}'

For IDEs that support direct MCP installation links, you can use a standardized deep link or JSON configuration like this example.

cursor://anysphere.cursor-deeplink/mcp/install?name=screenshot-website-fast&config=eyJzY3JlZW5zaG90LXdlYnNpdGUtZmFzdCI6eyJjb21tYW5kIjoibnB4IiwiYXJncyI6WyIteSIsIkBqdXN0LWV2ZXJ5L21jcC1zY3JlZW5zaG90LXdlYnNpdGUtZmFzdCJdfX0=

If you work within JetBrains IDEs, add the MCP server as a JSON snippet in the Model Context Protocol under As JSON with this command.

{
  "command":"npx",
  "args":["-y","@just-every/mcp-screenshot-website-fast"]
}

Raw JSON (works in any MCP client) also demonstrates the server configuration you can drop into your MCP config file.

{
  "mcpServers": {
    "screenshot-website-fast": {
      "command": "npx",
      "args": ["-y", "@just-every/mcp-screenshot-website-fast"]
    }
  }
}

Additional notes

This MCP server is designed for AI vision workflows with features like tiling, viewport configuration, and optional screencasts. It focuses on providing reliable, up-to-date captures by avoiding caching and enabling precise control over wait strategies and image sizes.

Troubleshooting

If you encounter Puppeteer-related issues, ensure Chrome/Chromium can be downloaded or provide a custom executable path. For screenshot quality, adjust the viewport or try different wait strategies. If you see timeout errors, increase the wait time or try alternate wait conditions.

Development notes

Development workflows use npm to install, build, and run tests. The project offers commands for taking full-page or viewport-only captures, with options for saving to a directory and waiting for specific load events.

Available tools

take_screenshot

Captures a high-quality screenshot of a webpage. Supports optional width, height, fullPage tiling, waitUntil, waitFor, and output directory.

take_screencast

Captures a series of screenshots over time to create a screencast. Supports duration, interval, JavaScript evaluation at start, wait strategy, and optional output directory for animated WebP.