中文 | Deutsch | Español | français | 日本語 | 한국어 | Português | Русский

Fetcher MCP

MCP server for fetch web page content using Playwright headless browser.

🌟 Recommended: OllaMan - Powerful Ollama AI Model Manager.

Advantages

• JavaScript Support: Unlike traditional web scrapers, Fetcher MCP uses Playwright to execute JavaScript, making it capable of handling dynamic web content and modern web applications.

• Intelligent Content Extraction: Built-in Readability algorithm automatically extracts the main content from web pages, removing ads, navigation, and other non-essential elements.

• Flexible Output Format: Supports both HTML and Markdown output formats, making it easy to integrate with various downstream applications.

• Parallel Processing: The fetch_urls tool enables concurrent fetching of multiple URLs, significantly improving efficiency for batch operations.

• Resource Optimization: Automatically blocks unnecessary resources (images, stylesheets, fonts, media) to reduce bandwidth usage and improve performance.

• Robust Error Handling: Comprehensive error handling and logging ensure reliable operation even when dealing with problematic web pages.

• Configurable Parameters: Fine-grained control over timeouts, content extraction, and output formatting to suit different use cases.

Quick Start

Run directly with npx:

npx -y fetcher-mcp

First time setup - install the required browser by running the following command in your terminal:

npx playwright install chromium

HTTP and SSE Transport

Use the --transport=http parameter to start both Streamable HTTP endpoint and SSE endpoint services simultaneously:

npx -y fetcher-mcp --log --transport=http --host=0.0.0.0 --port=3000

After startup, the server provides the following endpoints:

• /mcp - Streamable HTTP endpoint (modern MCP protocol)
• /sse - SSE endpoint (legacy MCP protocol)

Clients can choose which method to connect based on their needs.

Debug Mode

Run with the --debug option to show the browser window for debugging:

npx -y fetcher-mcp --debug

Configuration MCP

Configure this MCP server in Claude Desktop:

On MacOS: ~/Library/Application Support/Claude/claude_desktop_config.json

On Windows: %APPDATA%/Claude/claude_desktop_config.json

{
  "mcpServers": {
    "fetcher": {
      "command": "npx",
      "args": ["-y", "fetcher-mcp"]
    }
  }
}

Docker Deployment

Running with Docker

docker run -p 3000:3000 ghcr.io/jae-jae/fetcher-mcp:latest

Deploying with Docker Compose

Create a docker-compose.yml file:

version: "3.8"

services:
  fetcher-mcp:
    image: ghcr.io/jae-jae/fetcher-mcp:latest
    container_name: fetcher-mcp
    restart: unless-stopped
    ports:
      - "3000:3000"
    environment:
      - NODE_ENV=production
    # Using host network mode on Linux hosts can improve browser access efficiency
    # network_mode: "host"
    volumes:
      # For Playwright, may need to share certain system paths
      - /tmp:/tmp
    # Health check
    healthcheck:
      test: ["CMD", "wget", "--spider", "-q", "http://localhost:3000"]
      interval: 30s
      timeout: 10s
      retries: 3

Then run:

docker-compose up -d

Features

• fetch_url - Retrieve web page content from a specified URL

  • Uses Playwright headless browser to parse JavaScript
  • Supports intelligent extraction of main content and conversion to Markdown
  • Supports the following parameters:
    • url: The URL of the web page to fetch (required parameter)
    • timeout: Page loading timeout in milliseconds, default is 30000 (30 seconds)
    • waitUntil: Specifies when navigation is considered complete, options: 'load', 'domcontentloaded', 'networkidle', 'commit', default is 'load'
    • extractContent: Whether to intelligently extract the main content, default is true
    • maxLength: Maximum length of returned content (in characters), default is no limit
    • returnHtml: Whether to return HTML content instead of Markdown, default is false
    • waitForNavigation: Whether to wait for additional navigation after initial page load (useful for sites with anti-bot verification), default is false
    • navigationTimeout: Maximum time to wait for additional navigation in milliseconds, default is 10000 (10 seconds)
    • disableMedia: Whether to disable media resources (images, stylesheets, fonts, media), default is true
    • debug: Whether to enable debug mode (showing browser window), overrides the --debug command line flag if specified

• fetch_urls - Batch retrieve web page content from multiple URLs in parallel

  • Uses multi-tab parallel fetching for improved performance
  • Returns combined results with clear separation between webpages
  • Supports the following parameters:
    • urls: Array of URLs to fetch (required parameter)
    • Other parameters are the same as fetch_url

• browser_install - Install Playwright Chromium browser binary automatically

  • Installs required Chromium browser binary when not available
  • Automatically suggested when browser installation errors occur
  • Supports the following parameters:
    • withDeps: Install system dependencies required by Chromium browser, default is false
    • force: Force installation even if Chromium is already installed, default is false

Tips

Handling Special Website Scenarios

Dealing with Anti-Crawler Mechanisms

• Wait for Complete Loading: For websites using CAPTCHA, redirects, or other verification mechanisms, include in your prompt:

  Copy

  Please wait for the page to fully load
  

  This will use the waitForNavigation: true parameter.

• Increase Timeout Duration: For websites that load slowly:

  Copy

  Please set the page loading timeout to 60 seconds
  

  This adjusts both timeout and navigationTimeout parameters accordingly.

Content Retrieval Adjustments

• Preserve Original HTML Structure: When content extraction might fail:

  Copy

  Please preserve the original HTML content
  

  Sets extractContent: false and returnHtml: true.

• Fetch Complete Page Content: When extracted content is too limited:

  Copy

  Please fetch the complete webpage content instead of just the main content
  

  Sets extractContent: false.

• Return Content as HTML: When HTML format is needed instead of default Markdown:

  Copy

  Please return the content in HTML format
  

  Sets returnHtml: true.

Debugging and Authentication

Enabling Debug Mode

• Dynamic Debug Activation: To display the browser window during a specific fetch operation:
  Copy

  Please enable debug mode for this fetch operation
  

  This sets debug: true even if the server was started without the --debug flag.

Using Custom Cookies for Authentication

• Manual Login: To login using your own credentials:

  Copy

  Please run in debug mode so I can manually log in to the website
  

  Sets debug: true or uses the --debug flag, keeping the browser window open for manual login.

• Interacting with Debug Browser: When debug mode is enabled:

  1. The browser window remains open
  2. You can manually log into the website using your credentials
  3. After login is complete, content will be fetched with your authenticated session

• Enable Debug for Specific Requests: Even if the server is already running, you can enable debug mode for a specific request:

  Copy

  Please enable debug mode for this authentication step
  

  Sets debug: true for this specific request only, opening the browser window for manual login.

Development

Install Dependencies

npm install

Install Playwright Browser

Install the browsers needed for Playwright:

npm run install-browser

Build the Server

npm run build

Debugging

Use MCP Inspector for debugging:

npm run inspector

You can also enable visible browser mode for debugging:

node build/index.js --debug

Related Projects

• g-search-mcp: A powerful MCP server for Google search that enables parallel searching with multiple keywords simultaneously. Perfect for batch search operations and data collection.

License

Licensed under the MIT License