Integrations
Provides a website fetching tool using the MCP Python SDK, allowing AI agents to retrieve and process web content using a simple 'fetch' tool with a URL parameter.
Simple MCP Tool Server
A simple MCP server that exposes a website fetching tool using SSE transport.
Requirements
- Python 3.10 or higher (tested on Python 3.13)
Installation
MCP Python SDK Documentation
The MCP Python SDK documentation has been split into smaller files and organized in the docs/
directory. This structure makes it easier for AI agents to navigate and understand the SDK. The documentation covers:
- Core concepts (servers, resources, tools, etc.)
- Running MCP servers in different modes
- Examples and advanced usage
- And more!
Usage
The package provides a command-line interface (CLI) with several commands to manage the MCP server:
Starting the server
Start the server on the default port (7000) or specify a custom port:
Managing the server
The restart command will:
- Stop any existing server on the specified port
- Start a new server in the background
- Wait until the server is responsive
- Log output to server.log
CLI quick reference
Command | Purpose |
---|---|
start | Start the server |
stop | Stop the server |
check | Health-check |
restart | Stop & start |
Server Tools
The server exposes the following tools:
- fetch: Remote HTTP fetcher – give an absolute URL; returns page text.
url
: The URL of the website to fetch (required)
- search_docs: Semantic search across SDK documentation; returns top-k excerpts.
query
: Search phrase or question (required)k
: Number of top matches to return (optional, default = 3)
- get_content: Get the full local file for any match returned by
search_docs
.file
: Path relative to docs (required)
Development Setup
For development, install additional tools:
Use the Makefile for common tasks:
The test suite has a built-in 20-second timeout for all tests to prevent hanging, especially with SSE endpoints. For individual tests, a more strict timeout can be specified using the @pytest.mark.timeout(seconds)
decorator.
Semantic Search Index
For the search_docs tool, you can manually build or rebuild the vector index:
The index is built automatically on first tool use if it doesn't exist.
Project Architecture
Using with Cursor
This MCP server can be used with Cursor as a client. For setup:
- Run the server in a terminal:
- Configure Cursor by creating a
.cursor/mcp.json
file:
- Mention the server in your prompts when using Cursor
This server cannot be installed
remote-capable server
The server can be hosted and run remotely because it primarily relies on remote services or has no dependency on the local environment.
A server that provides a website fetching tool via SSE transport, allowing users to retrieve content from specified URLs.
Related MCP Servers
- AsecurityAlicenseAqualityAn advanced web browsing server enabling headless browser interactions via a secure API, providing features like navigation, content extraction, element interaction, and screenshot capture.Last updated -69PythonMIT License
- AsecurityAlicenseAqualityThis server facilitates interaction with cosense/Scrapbox projects, enabling users to retrieve, list, search, and create pages while supporting various query operations and secure access to private projects.Last updated -412TypeScriptMIT License
- AsecurityAlicenseAqualityThis server enables users to download entire websites and their assets for offline access, supporting configurable depth and concurrency settings.Last updated -14PythonMIT License
- AsecurityAlicenseAqualityA simple MCP server that facilitates website fetching through a configurable server platform using stdio or SSE transport, allowing integration with tools like Cursor for streamlined access.Last updated -223PythonMIT License