MCP Page Capture

What chasesaurabh/mcp-page-capture MCP server does

The chasesaurabh/mcp-page-capture MCP server opens web pages in headless Chromium and exposes two MCP tools: captureScreenshot and extractDom. The screenshot tool returns image content, including PNG output encoded as base64, while the DOM tool returns structured page content and can limit extraction to a CSS selector.

It is intended for pages that require browser rendering rather than simple HTTP retrieval. A request can target an arbitrary URL and optionally provide request headers. The server can also work with cookies, which supports examples such as capturing a page behind a session or bearer-token-based access.

How it works

The server starts a Node.js MCP process and validates each incoming request against its schemas. Puppeteer launches or reuses a Chromium instance, navigates to the requested URL, performs the requested actions, and returns structured results to the MCP client.

Screenshot workflows use six primary step types:

• viewport sets a device preset or explicit dimensions.
• wait pauses for a selector or duration.
• fill enters values into text fields, selects, checkboxes, or radio buttons.
• click activates an element and can wait for another selector.
• scroll moves to a selector or vertical position.
• screenshot captures the viewport, an element, or the complete page.

The server adds a screenshot step when one is omitted and can correct step ordering by moving viewport first and screenshot last. Login and search patterns are also documented as higher-level inputs that expand into multiple actions.

Setup and configuration

The chasesaurabh/mcp-page-capture MCP server requires Node.js 18 or newer and npm 9 or newer. Chromium download permissions are needed on the first run, and the runtime must be able to reach the requested URLs. Install it from npm with npx mcp-page-capture, or clone the repository and run npm install, npm run build, and npm start.

For local MCP client configuration, the repository shows a Node command that starts dist/cli.js. The path may need to be absolute depending on the working directory. Development mode is available through npm run dev. Docker build and run instructions are also provided for environments that need a packaged Chromium setup.

Tools and capabilities

captureScreenshot supports normal viewport captures, full-page images, element screenshots, device presets such as mobile and tablet profiles, interaction steps, custom headers, and cookie setup. extractDom can retrieve rendered content for a page or a selected CSS region. The schema exposes four top-level request areas: url, steps, headers, and validate.

The implementation includes automatic retries with exponential backoff, structured logging, defensive error handling, telemetry hooks, and pluggable local filesystem, S3, and memory storage backends. These operational features are described by the project, but their detailed configuration is not specified in the supplied material.

Limitations and notes

The server depends on a working Chromium installation or download and network access to every target URL. Browser actions may require selectors that match the target page. The README documents advanced steps beyond the six primary LLM-facing steps, but their full schemas are not included in the supplied material. No external API key is listed as a requirement.