ldoce

The Ldoce MCP Server is a Node.js/TypeScript service thatscrapes data from the Longman Dictionary (https://www.ldoceonline.com/dictionary/<word>) and returns a structured JSON object compliant with the Model Context Protocol (MCP). It extracts an introduction, related topics, dictionary entries (including pronunciation, senses, and examples), corpus examples, and etymology for a given word. The server is designed to be consumed by MCP-enabled agents (for example, Claude Desktop) and provides a standardized tool output that MCP clients can invoke to enrich conversations with dictionary data. To use it, deploy the server in your environment and point your MCP client at its exposed endpoint to request information about a specific word; the server will respond with a JSON payload containing the parsed dictionary data and metadata.

Typical usage flow:

• An MCP client sends a request for a word to the Ldoce MCP Server.
• The server fetches the word page from Longman Dictionary, parses relevant sections, and structures the data into a standardized JSON object.
• The client consumes the data (definitions, examples, pronunciation, etymology, and related topics) to support reasoning, language tasks, or user queries.
• The MCP SDK integration exposes additional tooling to invoke this server as a functional knowledge source within supported MCP ecosystems.

Capabilities you can rely on include: extraction of introduction text, listing of related topics, detailed dictionary entries with senses and pronunciation, curated corpus examples, and word origin data, all formatted for easy ingestion by AI agents.

• Prerequisites:

  • Node.js (version 16 or higher)
  • npm
  • Git

• Option 1: Installing via Smithery (automatic for Claude Desktop):

  1. Ensure you have Smithery CLI installed.
  2. Run the command: npx -y @smithery/cli install @edgardamasceno-dev/ldoce-mcp-server --client claude

• Option 2: Manual installation

  1. Clone the repository: git clone https://github.com/seuusuario/ldoce-mcp-server.git cd ldoce-mcp-server
  2. Install dependencies: npm install
  3. Build (if project uses TypeScript): npm run build
  4. Start the server (adjust the entry path if necessary): node dist/server.js # or node dist/index.js depending on build output

• Environment configuration (optional):

  • If the server requires API keys or configurations, create a .env file or set environment variables as needed by the app. Typical variables might include API_BASE_URL, TIMEOUT_MS, or LONGMAN_BASE_URL, depending on implementation.

• Verification:

  • Once running, send a test request to the MCP endpoint (as defined by your server’s routing) to verify that the response contains the expected JSON structure with introduction, related topics, entries, corpus examples, and etymology.

Tips and common considerations:

• The server relies on web scraping (Axios + Cheerio) to fetch and parse Longman Dictionary pages; if the source page structure changes, you may need to adjust selectors in the parsing layer.
• Ensure respect for the target site's robots.txt and terms of service when scraping data.
• Monitor request rate to avoid being blocked by the source website; implement caching if supported to reduce repeated fetches for the same word.
• If you plan to expose the MCP server publicly, secure endpoints and consider authentication/authorization for production deployments.
• The MCP JSON output should maintain consistent field structure across requests to maximize compatibility with MCP SDKs and clients.