Dictionary Lookup MCP Server

A minimal Model Context Protocol server built with the official Python SDK. It exposes one tool, define, which looks up the definition of an English word using the free dictionaryapi.dev API.

This project was built as a learning exercise to understand how MCP servers are structured, how they communicate with clients, and how to deploy one publicly.

What it does

• Exposes a single tool: define(word: str) -> str

• Given a word, returns its part of speech and definition

• No API key or authentication required

Related MCP server: Hello World MCP FastAPI Endpoint

Requirements

• Python 3.10+

• uv (recommended) or pip

Setup (local)

# 1. Clone this repo
git clone <your-repo-url>
cd my-mcp-server

# 2. Install dependencies
uv sync

Running it locally

Option A — MCP Inspector (visual testing tool, easiest for beginners):

uv run mcp dev server.py

This opens a browser-based inspector where you can call the define tool directly and see the request/response without needing a full AI client.

Option B — Connect to Claude Desktop:

Add this to your Claude Desktop config file (claude_desktop_config.json):

{
  "mcpServers": {
    "dictionary-lookup": {
      "command": "uv",
      "args": [
        "--directory",
        "/absolute/path/to/my-mcp-server",
        "run",
        "server.py"
      ]
    }
  }
}

Restart Claude Desktop, then ask something like: "Use the dictionary tool to define 'ephemeral'."

Usage example

Input: define("ephemeral") Output: ephemeral (adjective): lasting for a very short time.

Deployment

This server is published on Smithery: <link once deployed>

To deploy your own copy:

1. Push this repo to GitHub.

2. Ensure server.py supports the streamable-http transport for remote hosting.

3. Host it (e.g. Render/Railway) or use Smithery's publish flow.

4. On Smithery, choose "Publish via URL" and point it at your hosted server's endpoint.

Why this project

Built to learn the fundamentals of MCP: how a client discovers a server's tools, how a tool call is made and answered, and how a working server gets shared publicly through a marketplace like Smithery.