LabscoMCP early access ↗
seanivoreMCP SERVER · OFFICIAL PROJECT

MCP File Preview Server

Screenshot a local HTML file with its CSS applied, and count what is actually in it — headings, paragraphs, images, links.

by seanivore ↗Browser Automation, Control & Screenshots✓ Official

LABSCO SUMMARY

For the file on disk, not the page on the network.

Most screenshot servers want a URL. This one takes a path, which is the case that comes up when you have just generated an HTML file and want to know whether it looks right before shipping it. The structure counts are the cheaper half — they answer "did the headings survive" without pulling the whole document into context.

What it is

A two-tool server for looking at HTML files on disk. It renders the file in a real browser so local stylesheets and resources resolve, and it can report the document's structure without you reading the markup.

What you get

  • preview_file takes a filePath and captures a full-page screenshot with the file's CSS applied, returning the HTML content alongside it
  • Optional width and height control the viewport for that capture
  • Screenshots are written to a screenshots directory in the project folder rather than only being returned inline
  • analyze_content takes a filePath and returns counts of headings, paragraphs, images and links

Requirements

The npm package mcp-file-preview, version 1.0.0, built from a checkout with npm run build and launched as node build/index.js. It renders through Puppeteer, so a Chromium download happens on install. No account or key.

Setup effort

Build from source — git clone https://github.com/seanivore/mcp-file-preview.git