GeekNews

GeekNews MCP Server provides a Python-based MCP endpoint that scrapes GeekNews articles and weekly news, caches data to reduce load, and exposes tools for retrieving current articles and weekly summaries via MCP clients. It automates daily parsing and caching so you can query data quickly without hitting the source on every request.

How to use

You will run a MCP client that connects to the GeekNews MCP Server to fetch articles or weekly news. Use the available tools to request data like top, new, ask, or show articles, or to retrieve weekly news with item-level details. The server serves data from a cache when valid, automatically refreshing after 24 hours.

How to install

Prerequisites you need before installing this MCP server are Node.js and Python, plus the MCP client tooling used to run MCP servers locally.

# 1) Clone the project repository
git clone https://github.com/the0807/GeekNews-MCP-Server
cd GeekNews-MCP-Server

# 2) Set up the runtime environment with the UVX tool
uv sync

# 3) Create and activate a virtual environment
uv venv
source .venv/bin/activate

# 4) Run an MCP test using the MCP Inspector (example test command)
uv run mcp
mcp dev main.py

# Open the URL shown in the terminal (MCP Inspector) to test the server

Configuration and notes

Configuration is provided through the MCP runtime. The server supports a single MCP configuration that specifies how to launch the server and its API key. The example below demonstrates how to register the GeekNews MCP Server with a client using the Smithery CLI.

{
  "mcpServers": {
    "geeknews_mcp": {
      "command": "npx",
      "args": [
        "-y",
        "@smithery/cli@latest",
        "run",
        "@the0807/geeknews-mcp-server",
        "--key",
        "YOUR_KEY"
      ]
    }
  }
}

Caching and data flow

The server automatically parses GeekNews data once per day and stores it in a local cache. When cached data is valid (within 24 hours), the server serves responses from the cache to minimize load on the GeekNews website. A scheduler periodically checks the cache validity and refreshes data as needed.

Available tools

get_articles

Fetches articles from GeekNews with specified type (top, new, ask, show) and a defined count. Returns title, URL, points, author, time, comment count, and ranking, using cached data when available.

get_weekly_news

Fetches weekly GeekNews news by ID or the most recent week. Returns weekly title, number, ID, content, URL, and items; each item includes title, URL, and ranking, using cached data when available.