UXMachine or Playwright MCP: which one should your AI agent use to check a website?
Use Playwright MCP when your AI agent needs to drive a browser and interact with a site. Use UXMachine when you need the same checks on your published pages after every change, and to see which observations changed since the previous measurement. They do different jobs and can be used together.
What Playwright MCP does
- An MCP server from Microsoft that lets an AI agent control a browser through Playwright, using accessibility snapshots of the page.
- The agent can navigate, click, type, fill in forms, upload files, take screenshots and read console messages and network requests.
- It runs on your machine by default (npx), with a visible or headless browser. It can also run as an HTTP server or in Docker.
- It can emulate a phone, by device name or viewport size.
- Its documentation does not describe accessibility or UX reports, or a comparison between runs.
- Open source under the Apache 2.0 license.
What UXMachine does
- 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. Each observation says where and at which width it was seen, the state to reach and how to check it again.
- Reports what it could not check, and why.
- 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, and what could not be checked.
- It does not fill in forms, enter password-protected areas or follow what happens after a click. No AI model takes part in the check. A measurement uses 1 credit only when it returns a report with content.
When to use which
- Playwright MCP, when your agent has to interact with the site: try a sign-up flow, reproduce a bug, check something behind a login.
- UXMachine, when you want the same checks on the published pages after every change, and to know which observations changed since last time.
- Both: Playwright MCP to explore and interact with a flow; UXMachine afterwards to run the same checks and compare them with the previous measurement.
Do not use UXMachine if your agent needs to interact with the page: click, type, fill in forms, upload files, sign in, or read console and network activity. Use Playwright MCP for that.
This page compares documented features, not results on your site. UXMachine does not replace a full audit or testing with people. Checked on 19 October 2026.
Sources: Playwright MCP README