UXMUXMachine

Is there an MCP server to check a website for UX and accessibility problems?

Yes. UXMachine is an MCP server that checks a website in a real browser and returns what it observed, with evidence, and what it could not check. Your AI coding agent calls it like any other tool. Other MCP servers can drive a browser or run audits; they do different jobs, explained below.

What UXMachine checks

  • Opens up to 11 pages of a site in a browser running on UXMachine's servers, at 1280×800 and 390×844.
  • Runs a fixed, public list of checks. Among them: text contrast, images without alt text, blocked zoom, buttons and links without a name, a missing main heading, and pages without a title or a description.
  • Each observation says where and at which width it was seen, the state to reach and how to check it again. No AI model takes part in the check.
  • It reports what it could not check, and why. A short report is not the same as a clean site.
  • When there is a previous comparable measurement of the same site, it reports what is new, still observed, observed again, no longer observed or not comparable.

How your agent uses it

  • Connect the server at https://uxmachine.app/mcp in Claude, Claude Code, Cursor, Lovable or another MCP client. The steps for each are on the agents page.
  • The agent calls medir with the address, then estado until the report is ready, a few minutes later.
  • For a version running on your machine, the tunel tool lets the agent measure it before you publish.

Other MCP servers you may be considering

  • Playwright MCP lets your agent drive a browser: navigate, click, type and fill in forms. Its documentation does not describe accessibility or UX reports. See UXMachine or Playwright MCP?
  • Chrome DevTools MCP lets your agent inspect a live page, and its Lighthouse tool audits accessibility, SEO, best practices and agentic browsing. See UXMachine or Chrome DevTools MCP?

What it does not do

  • It does not fill in forms, enter password-protected areas or follow what happens after a click.
  • It does not measure speed or give a score.
  • It does not replace a full accessibility audit or testing with people.
  • A measurement uses 1 credit only when it returns a report with content.

This page describes documented features, not results on your site. Checked on 20 October 2026.

Sources: Playwright MCP README · Chrome DevTools MCP README · Lighthouse tool source