PlugWiki
mcp-playwright logo

ExecuteAutomation Playwright MCP Server: Setup for Claude, Cursor and Other Apps

CommunityLocal
Getnpm@executeautomation/playwright-mcp-serverexecuteautomation/mcp-playwright

Not the official Playwright MCP server. This one is built by the community. Playwright makes its own: Playwright MCP →

Lets Claude, Cursor and other assistants open a real browser, click and type on pages, take screenshots, run JavaScript and send API requests.

This is a community server by ExecuteAutomation, not the Playwright team's own. It runs on your computer and drives Chromium, Firefox or WebKit through Playwright. The browser opens in a visible window unless the assistant asks for headless mode. It needs no account or API key, and it downloads the browsers it needs the first time you use it.

It can run any JavaScript in the page, upload files from your computer to a site, and send POST, PUT and DELETE requests. Start it on test sites first. Microsoft publishes a separate server for Playwright, so check which one a guide means before you copy its config.

Good for

  • Developers who want the assistant to test a web app in a real browser
  • QA engineers who want Playwright test code written from the steps the assistant takes
  • Anyone who wants to scrape or check a page as it looks on a phone or tablet
  • Teams that need quick API checks next to the browser steps

Works in

12 should work

Tap an app to see how to set it up and any known problems.

Install mcp-playwright

Claude Code

  1. 1

    Copy this prompt.

    Add the mcp-playwright server to Claude Code. Run this command:
    
    claude mcp add playwright -- npx -y @executeautomation/playwright-mcp-server
    
    Then check that it loaded and tell me if I need to sign in or restart.
  2. 2

    Paste it into Claude Code. It shows the change and asks before running it.

Requires Node.js.

Built from the package data we hold. Not yet tested by us.

Tools

  • Start recording actions as a test

    Starts a session that records the browser steps your assistant takes, so they can become a Playwright test file. You give it a folder to save into.

    Try: “Start recording a test into /Users/me/tests.”

    Write · start_codegen_session

  • Finish recording and write the test file

    Ends a recording session and writes the generated test file into the folder you chose.

    Try: “End the recording and save the test.”

    Write · end_codegen_session

  • Look at a recording session

    Returns details about one recording session by its ID.

    Try: “Show me the current recording session.”

    Read · get_codegen_session

  • Throw away a recording

    Clears a recording session without writing a test.

    Try: “Discard the recording.”

    Write · clear_codegen_session

  • Open a web address in the browser

    Opens a URL in a Chromium, Firefox or WebKit browser. A visible browser window opens unless you ask for headless mode.

    Try: “Open https://example.com in a browser.”

    Network · playwright_navigate

  • Take a screenshot

    Takes a screenshot of the page or of one element, and can save it as a PNG file.

    Try: “Take a screenshot of the pricing table.”

    Files · playwright_screenshot

  • Click an element

    Clicks an element on the page that you pick with a CSS selector.

    Try: “Click the Sign in button.”

    Write · playwright_click

  • Click inside an iframe

    Clicks an element that sits inside an iframe on the page.

    Try: “Click Pay inside the payment frame.”

    Write · playwright_iframe_click

  • Type into a field inside an iframe

    Types a value into an input that sits inside an iframe.

    Try: “Type the email address into the checkout frame.”

    Write · playwright_iframe_fill

  • Fill in an input field

    Types a value into an input field you pick with a CSS selector.

    Try: “Fill the search box with 'running shoes'.”

    Write · playwright_fill

  • Pick a dropdown option

    Chooses an option in a select dropdown.

    Try: “Select Canada in the country dropdown.”

    Write · playwright_select

  • Hover over an element

    Moves the mouse over an element, for menus that open on hover.

    Try: “Hover over the Products menu.”

    Write · playwright_hover

Read from the source code

Permissions

high risk · 5 things to watch

High risk

The assistant can drive a real browser, run its own JavaScript on pages, upload files and send API requests that change or delete data. Use it on test sites with test accounts and watch the browser window while it works.

What it can do

  • Read your data
  • Create and change things
  • Delete things
  • Read-only mode

Watch out for

  • It runs JavaScript on any page it opens

    The evaluate tool runs code you or the assistant provide inside the browser page. Only use it on sites you trust, and never while signed in to accounts you care about. github.com

  • It can send DELETE, PUT and POST requests

    The HTTP tools act with whatever token the assistant passes in. Give it test credentials, not production tokens. github.com

  • It can upload local files to a website

    The upload tool takes any absolute file path on your computer and puts that file into a page's upload field. Keep sensitive files out of reach and watch which page is open. github.com

  • It writes files to folders it is given

    Screenshots, PDFs and generated tests are saved to paths the assistant chooses. Check the folder before you accept. github.com

  • Local HTTP mode has no sign-in

    The optional port mode is documented as localhost only. For remote use the README suggests an SSH tunnel instead of opening it up. github.com

Security advisories

No published security advisories as of Nov 1, 2026.

Technical details
Turning on read-onlyNo read-only mode is documented.
Talks toAny website or API address the assistant opens or calls from your computer
Sign-in and keysNo account or key. Optional bearer tokens are passed in individual tool calls.

Questions

Is this the official Playwright MCP server?

No. It is made by ExecuteAutomation, a community project. Microsoft publishes its own Playwright server. The two have different tools and different install commands.

Does it need an API key?

No. It needs no account or key. It downloads the Chromium, Firefox and WebKit browsers on first use, or you can install them yourself with the Playwright command in the README.

Does it work in Claude Desktop?

Yes, in standard stdio mode, which the README recommends for Claude Desktop. The optional port mode is for VS Code, custom clients and remote use, and needs type http in the client config.

Can it run without a visible browser window?

Yes. The browser opens in a visible window by default, and the navigate tool takes a headless option to hide it.

Can I run it read-only?

No. No read-only mode is documented. It has click, fill, upload, JavaScript and write-style HTTP tools, so only point it at sites and accounts you can afford to change.

Can it write Playwright tests?

Yes. The recording tools save the steps the assistant performs as a Playwright test file in a folder you choose.

Is it free?

Yes. It is open source under the MIT license and runs on your computer with no usage fee.

Information

updated Nov 2, 2026

Publisher
Community
Package
@executeautomation/playwright-mcp-server
Registry
npm
Transport
stdio
Auth
None
License
MIT
Last commit
Dec 13, 2025
Open issues
32
Listed in
Docker MCP Catalog

WebsiteGitHubDockerDocker MCP CatalogIssues

Other Playwright MCP servers

Sources

Docs, issue trackers and discussions this page is built from. Issues, quotes and findings above link to their own source.

Show all 2
  1. github.com · github.com
  2. github.com · github.com

Facts from the official MCP registry, GitHub, npm, PyPI and the Docker catalog. Updated Nov 2, 2026.