Fetch

What it is

Fetch MCP Server is a Model Context Protocol server that lets LLMs retrieve and process content from web pages. It converts HTML to markdown for easier consumption. It exposes a fetch tool and a fetch prompt, and truncates responses, letting models read long pages in chunks through the start_index argument.

Who it's for

• Developers who want to give an MCP-capable LLM client (such as Claude.app or VS Code) the ability to read web pages
• Users who need long web pages read in chunks until the relevant information is found

Requirements

Setup

1. Run with uvx (recommended)

   With uv, no specific installation is needed; uvx runs mcp-server-fetch directly. Add this to your Claude settings.

   jsonCopy

   {
     "mcpServers": {
       "fetch": {
         "command": "uvx",
         "args": ["mcp-server-fetch"]
       }
     }
   }

2. Install with pip

   Alternatively, install mcp-server-fetch via pip.

   bashCopy

   pip install mcp-server-fetch

3. Run the pip-installed server as a script

   After installing with pip, run it as a module.

   bashCopy

   python -m mcp_server_fetch

4. Configure with Docker

   Add this to your Claude settings to run the server via Docker.

   jsonCopy

   {
     "mcpServers": {
       "fetch": {
         "command": "docker",
         "args": ["run", "-i", "--rm", "mcp/fetch"]
       }
     }
   }

5. Configure for VS Code (uvx)

   Add this JSON block to your User Settings (JSON) file, or to .vscode/mcp.json in your workspace to share it with others. The mcp key is needed when using the mcp.json file.

   jsonCopy

   {
     "mcp": {
       "servers": {
         "fetch": {
           "command": "uvx",
           "args": ["mcp-server-fetch"]
         }
       }
     }
   }

Examples

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

npx @modelcontextprotocol/inspector uvx mcp-server-fetch

fetch(url=<URL>, max_length=5000, start_index=0, raw=false)

Pros & cons

Images