MCP Web Tools

Provides web search and fetch capabilities via a local MCP server for AI assistants, enabling web requests from your machine.
  • python

0

GitHub Stars

python

Language

5 months ago

First Indexed

2 months ago

Catalog Refreshed

Documentation & install

Readme and setup notes from the catalogue, plus a client-ready config you can copy for your MCP host.

Installation

Add the following to your MCP client configuration file.

Configuration

View docs
{
  "mcpServers": {
    "sayreblades-mcp-web-tools": {
      "command": "uv",
      "args": [
        "run",
        "mcp-web-tools"
      ]
    }
  }
}

You can run MCP Web Tools locally to perform web searches and fetch web content directly from your machine, enabling web requests that originate inside your own network and are not proxied through external servers.

How to use

Start the MCP Web Tools server locally so your AI assistant can make web requests from your environment. You will interact with the server through an MCP client that supports stdio-based MCP servers. With the server running, your client can perform web searches and fetch operations using the two built-in endpoints: web_search and web_fetch. Use these capabilities to look up information on the web and to retrieve content from specific pages, with results formatted for easy consumption by your assistant.

How to install

Prerequisites you need on your machine before installing the MCP Web Tools server are Python 3.13 or newer and uv.

# Quick Install (Recommended)
# Clone the repository
git clone <repo-url>
cd mcp-web-tools

# Install globally and configure Claude Code
make install

If you prefer manual installation, you can run the server directly from the project with uv.

# Install dependencies
uv sync

# Run from the project directory
uv run mcp-web-tools

Configuration and runtime details

To run the server locally as an MCP stdio endpoint, start the server using the runtime command shown during installation. The final start command is the stdio entry you use to launch the server, which your MCP client will reference. For development and testing, you typically start the server with the following command in the project directory.

uv run mcp-web-tools

Available tools

web_search

Search the web using DuckDuckGo. Returns results with titles, URLs, and snippets, with configurable max results and region.

web_fetch

Fetch and extract content from web pages. Converts HTML to text, cleans scripts and navigation, and returns a JSON-formatted response with a configurable timeout.

Built by
VeilStrat
AI signals for GTM teams
© 2026 VeilStrat. All rights reserved.All systems operational