M
MCP Directory CURATED DIRECTORY
911 MCP servers indexed
中文
← Back to directory
F

Fetcher MCP Server

Community
MCP server for fetching web page content using Playwright headless browser.
How to install ↓ Compare GitHub source repository ↗
Category
Browser #25 of 32
Stars
★ 1.1k Popular
Transport
stdio (local process) · SSE · Streamable HTTP
Runtime
Node.js
Credentials
No credential needed
License
MIT
Last commit
⚠
Tools
3
43FMRS · D

Fetcher MCP is a powerful web scraping server that leverages Playwright to support dynamic content, and offers intelligent extraction, batch fetching, and other features. It is suitable for developers dealing with modern web applications, but note the anti-crawler mechanisms and resource consumption.

Strongest · Documentation 12/20 Weakest · Reliability 6/20

Reliability
6/20
Security and permissions
10/20
Maintenance
8/20
Documentation
12/20
Setup experience
7/20
Why each score
Reliability 6/20
Evidence: The repository includes full TypeScript source code implementing tools fetch_url, fetch_urls, and browser_install, using Playwright for page fetching. The README provides clear usage examples. However, no verifiable execution evidence (such as CI workflows and committed tests) is present, and some feature claims (parallel processing, resource optimization) are only partially confirmed in the code. Thus, uncertainty about delivery lowers the score.
Security and permissions 10/20
Evidence: No obvious malicious behavior, credential theft, or dangerous default actions are found. The server receives URLs and accesses external networks via Playwright, which is its intended function. No explicit authentication mechanism is present, but that may be unnecessary as the server is typically run locally. However, no clear input validation or sanitization for user-supplied URLs is documented, and dangerous operations (e.g., browser installation) lack confirmation. These risks are visible but not fully mitigated, hence score 10.
Maintenance 8/20
Evidence: The repository has recent commits (based on stars and open issues, but no specific dates provided) and regular releases. License is MIT and clear. However, no security response channel or contribution guidelines are provided. Dependency update status is unclear. Thus, maintenance is active but governance is incomplete, giving 8.
Documentation 12/20
Evidence: The README provides detailed installation, configuration examples (Claude Desktop, Docker), tool parameters, usage tips (e.g., handling anti-crawler, debugging, authentication). It also mentions related projects. But it lacks troubleshooting guides, limitations (e.g., site compatibility issues), cost information (e.g., resource usage), and contribution guidelines. For instance, how to handle HTTP errors or network issues is not addressed. Therefore, documentation is usable but incomplete, giving 12.
Setup experience 7/20
Evidence: The README offers quick start steps via npx or Docker, with a Claude Desktop configuration example. Steps are clear, but require manual installation of Playwright browser (though there is a browser_install tool). Client configuration example is only for Claude Desktop, lacking other mainstream clients (e.g., Cursor, VS Code). The install steps rely on unverified npx commands, which may be fragile. Thus, score 7.

Static review · not runListed 2026-08-07

Read the FMRS scoring method →

Fit and risk

What it can accessUses the networkControls a browser

Best for

  • Developers needing to scrape modern JavaScript-rendered websites
  • Data collection scenarios requiring batch fetching
  • Users who want automatic extraction of main content

Not for

  • Scenarios where full original HTML is required and content extraction is prohibited (can be bypassed with settings, but extraction is default)
  • Users who need persistent login sessions but cannot provide custom cookies
  • Deployment in environments without Node.js and without desire to use Docker

Required permissions

  • Access to network to download web resources
  • Running a headless browser (Playwright Chromium)
  • Read and write temporary files (for browser cache)

Risks and side effects

  • May be identified as a crawler by target websites and trigger anti-bot mechanisms
  • Headless browser execution may consume significant memory and CPU
  • Fetched content may be subject to copyright or terms of use restrictions

Setup

Before you start

Runtime：Node.js

  1. Run directly with npx: npx -y fetcher-mcp.
  2. First-time setup: install the required browser by running npx playwright install chromium.
  3. For HTTP and SSE transport, use --transport=http --host=0.0.0.0 --port=3000.
  4. Configure MCP in Claude Desktop by editing the config file (see above).
claude_desktop_config.json 
{
  "mcpServers": {
    "fetcher": {
      "command": "npx",
      "args": [
        "-y",
        "fetcher-mcp"
      ]
    }
  }
}

Shown for Claude Desktop. Other clients may use a different file or key (VS Code uses "servers") — the configurator below converts it.

.vscode/mcp.json
{
  "servers": {
    "fetcher": {
      "command": "npx",
      "args": [
        "-y",
        "fetcher-mcp"
      ]
    }
  }
}

Goes in your project's .vscode/mcp.json (VS Code uses a "servers" key).

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

Run it in a terminal; replace any <…> placeholders with your own values first.

Generate and validate client config

Check that it works

The client's tool list should show fetch_url, fetch_urls, and browser_install; asking the model to fetch a simple page's content confirms the connection. Run npx playwright install chromium first if the browser is not installed.

Troubleshooting

  1. If you get a browser not installed error, run `npx playwright install chromium`
  2. If the page loading times out, increase the `timeout` parameter or set `waitUntil` to 'networkidle'
  3. If content extraction is incomplete, try setting `extractContent: false` and `returnHtml: true` to get the raw HTML

Things to try

Once connected, you can ask your AI assistant things like:

  • Fetch the content of https://example.com and convert it to Markdown
  • Fetch the main content of these three URLs in parallel
  • Please wait for the page to fully load and fetch this site with anti-bot verification
  • Return the complete webpage content as HTML instead of just the extracted main content

Tools 3

fetch_url read-only
Fetches web page content from a specified URL, with support for JavaScript rendering, intelligent content extraction, Markdown conversion, timeout and content extraction settings.
fetch_urls read-only
Fetches content from multiple URLs in parallel, improving batch processing efficiency.
browser_install writes
Automatically installs the Playwright Chromium browser binary, with options to install system dependencies and force installation.

Use cases

Fetching dynamic web pages that render JavaScript
Batch fetching multiple web pages for data collection
Converting web content to Markdown for integration into documents or knowledge bases
Handling anti-crawler mechanisms such as CAPTCHA or redirects

Supported clients

Claude Desktop

Listed from the project's documentation, not tested by this site.

Overview

Fetcher MCP is an MCP server that uses Playwright headless browser to fetch web page content. It supports JavaScript execution, enabling it to handle dynamic web content and modern web applications. Built-in Readability algorithm automatically extracts the main content from web pages, removing ads, navigation, and other non-essential elements. Supports both HTML and Markdown output formats. The fetch_urls tool enables concurrent fetching of multiple URLs, improving efficiency for batch operations. Automatically blocks unnecessary resources (images, stylesheets, fonts, media) to reduce bandwidth usage. Includes robust error handling and logging. Configurable parameters for timeouts, content extraction, and output formatting.

Similar servers

Compare with other MCP servers

Source revision 8754aff66e3d Data synced 2026-10-11 Read the FMRS scoring method