PlugWiki

Fetcher MCP Server: Read JavaScript Web Pages from Claude

CommunityLocal
Getnpmfetcher-mcpjae-jae/fetcher-mcp
Tools
3
2 read · 1 change
Downloads
12.1K
a week
Works in
12
apps
Risk
Medium
can write
Auth
None

Lets Claude, Cursor and other assistants open web pages in a real headless browser and read the main content as Markdown or HTML.

This is a community server by jae-jae. It runs on your computer with Playwright and a Chromium browser, and it needs no account or key. The first run needs the browser installed, either with the Playwright command in the README or by asking the assistant to use the browser_install tool.

It can also run as a web service for several clients, and as a Docker container. The README supports the plain local setup for Claude Desktop, and the install box covers other apps.

Good for

  • People whose assistant cannot read pages that need JavaScript to show their content
  • Researchers who want several pages fetched at once and cleaned of ads and menus
  • Developers who need to log in by hand once and then let the assistant read the page

Works in

12 should work

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

Install Fetcher MCP

Claude Code

  1. 1

    Copy this prompt.

    Add the Fetcher MCP server to Claude Code. Run this command:

claude mcp add fetcher -- npx -y fetcher-mcp

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

  • Opens a web page in a headless browser and returns its text

    Loads one URL in a headless Chromium browser so JavaScript runs, pulls out the main article with a readability filter, and returns Markdown or HTML. You can set a timeout, a length limit, and whether to wait for extra redirects or show the browser window.

    Try: “Fetch https://example.com/pricing and summarise the plans”

    Network · fetch_url

  • Opens several web pages at once

    Does the same as fetch_url for a list of URLs in parallel tabs and returns the results one after another with clear separators.

    Try: “Fetch these three docs pages and compare their setup steps”

    Network · fetch_urls

  • Downloads the Chromium browser it needs

    Runs the Playwright installer to fetch the Chromium browser, with options to also install system libraries or to reinstall. The server suggests it when a fetch fails because the browser is missing.

    Try: “Install the browser Fetcher needs”

    Runs code · browser_install

Read from the source code

Permissions

medium risk · 5 things to watch

Medium risk

The assistant can open web pages with a real browser and, if you allow it, install Chromium. Pages can contain instructions aimed at the assistant, so review what it reads and does next.

What it can do

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

Watch out for

  • It can install software on your computer

    browser_install runs the Playwright installer to download Chromium, and can also install system libraries. Install the browser yourself and decline the tool call if you do not want that. unpkg.com

  • It opens any address the assistant gives it

    The browser runs on your machine, so it can reach any address your computer can, including pages on your own network. Page text is also passed to the assistant, so a page can contain instructions aimed at it. Review what it fetches. unpkg.com

  • Web mode defaults to localhost

    With the HTTP option, the code binds to localhost unless you pass --host. The README's example uses --host=0.0.0.0, which exposes the service to your network with no sign-in. github.com

  • Installing the package downloads Chromium

    The package.json of fetcher-mcp 0.3.9 has a postinstall script that runs playwright install chromium, so the first npx run downloads a Chromium build from Playwright's servers. The README also tells you to run that command yourself on first setup. unpkg.com

  • The license is stated two ways

    The package.json in the npm package says ISC. The README, the LICENSE file in the repository and in the npm package say MIT (the file carries the copyright line of an unrelated author). Both are permissive; check the file you rely on. unpkg.com

Security advisories

No published security advisories as of Nov 1, 2026.

Technical details
Turning on read-onlyNo read-only switch is documented. fetch_url and fetch_urls only read pages, but browser_install is always available.
Talks toAny web address the assistant asks it to open. Playwright's download servers when it installs Chromium
Sign-in and keysNo credentials are needed. In debug mode you can sign in to a site by hand in the visible browser window, and the fetch then uses that session.

Questions

Is Fetcher MCP official?

No. It is a community project by jae-jae. It is not made by Playwright, Microsoft or any website it reads.

Does it need an API key?

No. It opens pages in a browser on your computer, so there is no key or account.

Why does it say the browser is missing?

The Chromium browser is not installed yet. The package tries to download it when you install it (a postinstall step), and the README also asks you to run npx playwright install chromium once. If a fetch still fails, ask the assistant to call browser_install.

Can it read pages that need a login?

Yes, with debug mode. It opens a visible browser window, you log in by hand, and it fetches the page with that session. You can also ask for debug mode on one request.

Does it run locally?

Yes. The default setup runs on your computer and talks to the app over stdio. It also has an HTTP mode and a Docker image for sharing one server.

Can it write or delete files?

It does not write files for you, apart from the Chromium download that browser_install triggers. The fetch tools only read pages.

How is it different from a normal fetch tool?

It runs JavaScript in a headless browser and extracts the main content, so single-page apps work. It also blocks images, styles, fonts and media by default to save bandwidth.

Which license does it use, ISC or MIT?

Both appear. The package.json in the npm release says ISC, while the README and the LICENSE file say MIT. Both are permissive, but the package metadata and the repository do not agree.

Information

updated Nov 2, 2026

Publisher
Community
Package
fetcher-mcp
Registry
npm
Transport
stdio, Streamable HTTP, SSE
Auth
None
Version
0.3.9
License
ISC

WebsiteGitHubnpmIssues

Sources

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

Show all 6
  1. Fetcher MCP README · github.com
  2. fetcher-mcp on npm · npmjs.com
  3. fetcher-mcp 0.3.9 fetchUrl.js · unpkg.com
  4. fetcher-mcp 0.3.9 browserInstall.js · unpkg.com
  5. fetcher-mcp 0.3.9 package.json · unpkg.com
  6. fetcher-mcp 0.3.9 LICENSE · unpkg.com

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