Home / MCP / Fetch MCP Server

Fetch MCP Server

Provides web content fetching and HTML-to-markdown conversion to enable chunked web content access for MCP clients.

other

74.1kstars

View on GitHub

Installation

Add the following to your MCP client configuration file.

Configuration

View docs

{
    "mcpServers": {
        "fetch": {
            "command": "uvx",
            "args": [
                "mcp-server-fetch"
            ],
            "env": {
                "PYTHONIOENCODING": "utf-8"
            }
        }
    }
}

You can use the Fetch MCP Server to retrieve web content and convert HTML to markdown, enabling language models to read and process online pages in manageable chunks. This is useful when you need structured web content as plain text and markdown for downstream tasks.

How to use

To use the Fetch MCP Server, run it locally or remotely and send it a URL to fetch. The server will return the page content converted to markdown, with support for chunked reading using a start_index to begin extraction at a specific character offset. This allows models to read web pages piece by piece until they locate the information they need. Use the fetch tool to request a URL and receive the extracted markdown content.

How to install

Prerequisites: you need a runtime to run the server. You can choose among several installation methods depending on your preference and environment.

Option 1: Using uvx (recommended)

Run the MCP server using uvx directly. No special installation is required beyond having uvx available.

Now install and run using uvx

# When using uv (via uvx) you run the server with:
# Command shown in the setup section
uvx mcp-server-fetch

Install via pip

pip install mcp-server-fetch

Run after pip install

python -m mcp_server_fetch

Separate install methods (examples in configuration)

The server can be configured in multiple ways depending on how you want to run it. Below are representative configuration blocks you might apply in your MCP client settings.

Available tools

fetch

Fetches a URL and extracts its contents as markdown for easy processing by models.