Provides web content fetching with browser automation, OCR, and multi-method extraction for MCP workflows.
Configuration
View docs{
"mcpServers": {
"maartensmeets-mcp-server-fetch": {
"command": "docker",
"args": [
"run",
"--rm",
"-i",
"mcp-server-fetch"
]
}
}
}Fetch MCP Server provides web content fetching capabilities through browser automation, OCR, and multiple extraction methods. This server empowers you to retrieve and process content from web pages, including those that require JavaScript rendering or advanced anti-scraping techniques, to feed into your MCP workflows.
You use the Fetch MCP Server by integrating it as an MCP tool in your client setup. The server exposes a fetch capability that retrieves a URL using browser automation and multiple extraction methods, then returns the content in a structured format suitable for processing by your LLM or downstream steps.
Prerequisites: you need Docker installed to run the provided container image. Optionally, you can build and run the server locally from source if you have the project available in your environment.
Install and run with Docker as a quick start:
docker build -t mcp-server-fetch .
docker run --rm -i mcp-server-fetchConfiguration, security, and usage notes are included to help you tailor the server to your setup. You can customize the user agent presented to downstream clients or models, and you can leverage the built-in cookie consent handling, full-page screenshots, and OCR-based extraction for content that is difficult to fetch.
Configure the MCP client to connect to the Fetch server as described below. When running in a client like Claude, you specify how to start the server process and how commands are routed.
{
"mcpServers": {
"fetch": {
"command": "docker",
"args": [
"run",
"--rm",
"-i",
"mcp-server-fetch"
],
"disabled": false,
"alwaysAllow": []
}
}
}If you encounter issues, enable debug logging in your MCP client to review scoring decisions and error indicators produced by the Fetch server. The server prioritizes content by a scoring system that favors longer, well-structured content and reduces the score for error messages or unreadable results.
Use the server in trusted environments and control access to the Docker image or container to prevent misuse. Review the user-agent customization to ensure you present a clear identity to target websites and downstream clients.
Fetches a URL from the internet using browser automation and multi-method extraction, including OCR. Returns content as markdown with support for JS-rendered pages, OCR-based extraction, HTML parsing, and document parsing.