ExecuteAutomation Playwright MCP

At a glance

This is the Playwright server from ExecuteAutomation, a test-automation training site, and it predates Microsoft's own Playwright MCP. The model drives a real Chromium, Firefox or WebKit browser by CSS selector: open pages, click, fill forms, run JavaScript, save PDFs and take screenshots. It can also record what it did as Playwright test code and make plain HTTP requests for API checks.

Its party trick is device emulation: one call switches the page to any of 143 Playwright device profiles, such as an iPhone 13 or an iPad in landscape. The last release was in December 2025, so for day-to-day browsing Microsoft's server is the safer default; this one suits people who want test code and device checks out of the session.

Made by

ExecuteAutomation (Karthik KK)

Licence

MIT

Latest version

1.0.12 (12 Dec 2025)

npm package

@executeautomation/playwright-mcp-server

GitHub stars

5.7K

Downloads

7K a week

Last commit

13 Dec 2025

Tools

33

Runs on

Your computer

Our check: We ran this exact config on 27 Oct 2026: it started and answered with 33 tools. The shared database listed "@michaellatman/mcp-get", an installer, as the package. The server itself is @executeautomation/playwright-mcp-server, version 1.0.12.

What's in the pod: its tools

We found 33 tools when we started it on 27 Oct 2026. 1 only read. The rest can change things, so keep your app's "ask before each tool call" setting on for them.

Grouped and described by us. The README on GitHub lists the inputs each tool takes.

Set it up

Pick your app for the exact snippet and where it goes.

{
  "mcpServers": {
    "executeautomation-playwright": {
      "command": "npx",
      "args": [
        "-y",
        "@executeautomation/playwright-mcp-server"
      ]
    }
  }
}

claude mcp add-json executeautomation-playwright '{"command":"npx","args":["-y","@executeautomation/playwright-mcp-server"]}'

{
  "mcpServers": {
    "executeautomation-playwright": {
      "command": "npx",
      "args": [
        "-y",
        "@executeautomation/playwright-mcp-server"
      ]
    }
  }
}

{
  "servers": {
    "executeautomation-playwright": {
      "type": "stdio",
      "command": "npx",
      "args": [
        "-y",
        "@executeautomation/playwright-mcp-server"
      ]
    }
  }
}

{
  "mcpServers": {
    "executeautomation-playwright": {
      "command": "npx",
      "args": [
        "-y",
        "@executeautomation/playwright-mcp-server"
      ]
    }
  }
}

For a client that connects by URL, start it with npx @executeautomation/playwright-mcp-server --port 8931 and point the client at http://localhost:8931/mcp.

New to this? See the setup guides for Claude Code, Cursor, Claude Desktop and ChatGPT.

Settings

Names are exactly as the project spells them. Required ones are marked.

What it can touch

Safer setup: It starts a fresh browser profile, so it is not signed in to your accounts unless you sign in during the session. Keep the HTTP tools away from production APIs.

Before you start

• Node.js 18 or newer
• Disk space for the Playwright browser it downloads on first use

If it does not work

The first call takes a long time

On first use it downloads the browser it needs. Run npx playwright install chromium beforehand to get that out of the way.

A headed browser fails on a server or in a container

There is no display. Ask for headless mode, or run the server in HTTP mode on a machine with a screen.

HTTP mode does not connect

Clients need "type": "http" alongside the URL http://localhost:8931/mcp. Claude Desktop only supports the stdio config.

No logs appear in the terminal

In stdio mode it writes logs to ~/playwright-mcp-server.log so they do not break the protocol.

Peas in a pod: similar servers

How ExecuteAutomation Playwright MCP lines up against other browser and web data servers on the shelf.

Questions

Something wrong on this page? Tell us.