Layout Doctor MCP server

Layout Doctor is a small helper that lets your AI actually look at an HTML page the way a browser would. You give it a file, and it tells you in plain numbers what is wrong: text sitting on top of other text, things sticking out of the screen, or leftover words like undefined. It is handy if you or your AI generate HTML pages or slides and you want to know whether they really look okay.

What this MCP server does

You point your AI at an HTML file, or paste the HTML, and tell it the screen size to use. Your AI passes that to this helper. The helper opens the page in a real browser (Chromium) behind the scenes and measures where every element and every piece of text actually lands. It then compares those positions and reports problems like overlapping text, elements outside the screen, or placeholders that were never filled in. You get a short report back in the chat, with the exact elements and pixel sizes involved.

Click to zoom

What you can do with it

• Check an HTML file for text that overlaps and becomes unreadable
• Find elements that stick out past the edge of the screen
• Spot leftover placeholders like undefined, NaN, or {{name}}
• Detect images, CSS, or scripts that point to files that do not exist
• See when content is clipped by overflow hidden and cannot be read
• Catch duplicate id values and invisible text
• Run the same check twice and get the same numbers both times

Try asking your AI

• “Check /Users/me/work/report.html at 1280 by 720 and tell me what is broken”
• “Here is my HTML for a pricing slide. Does any text overlap at 1280 by 800?”
• “Look at this page and list every placeholder that was not filled in”
• “Is anything sticking off the right side of the screen on this file?”

What it gives back to you

You get a short text report in the chat. It starts with a count of errors, warnings, and notes, then lists each problem with a short code, a plain sentence about what is wrong, the elements involved with their pixel positions, and a suggested fix. If you ask for JSON instead, it returns the same findings in a structured form your AI can work with.

Before you start

Who it's for

People who generate HTML pages or slides with AI and want a quick, honest check that the result actually looks right on screen.