home / mcp / levelang mcp server

Levelang MCP Server

Exposes levelang.app translation APIs via MCP with level-aware translation and multi-language support.

Installation

Add the following to your MCP client configuration file.

Configuration

{
  "mcpServers": {
    "beverage-levelang-mcp": {
      "url": "http://localhost:8463/mcp",
      "headers": {
        "MCP_PORT": "8463",
        "MCP_API_KEYS": "key1,key2",
        "MCP_TRANSPORT": "stdio",
        "LEVELANG_API_KEY": "YOUR_SERVICE_KEY",
        "LEVELANG_API_BASE_URL": "http://localhost:8000/api/v1"
      }
    }
  }
}

You run an MCP server that translates MCP tool calls into HTTP requests to the Levelang backend, exposing the levelang.app translation API to AI assistants while enforcing learner-friendly proficiency levels. This server is stateless, supports multiple languages and styles, and can be connected to MCP clients through either a live HTTP endpoint or a local stdio workflow.

How to use

Interact with the Levelang MCP server through your preferred MCP client (Claude Desktop, Cursor, Claude Code, or any MCP-compatible frontend). You can ask your assistant to translate text at a chosen learner level, discover supported languages and levels, or compare translations across levels and moods. Use the provided connection methods to reach the MCP server: a remote HTTP endpoint or a local stdio workflow via UV.

How to install

Prerequisites: Python 3.12 or newer, UV, and a running Levelang backend (local or remote). Follow these steps to install and run the MCP server.

# 1) Install dependencies and clone the MCP server repository
# Replace with the actual repository URL you were given
# git clone https://github.com/beverage/levelang-mcp.git
# cd levelang-mcp

# 2) Install Python dependencies if needed (UV is required for running the server)
# Typically handled by uv sync in the project setup

# 3) Start the MCP server shell for local development (stdio/workflow)
uv sync

# 4) Run the MCP server module directly if running in a development environment
# This depends on your setup; use the equivalent command provided below in the stdio example

Additional content

Configuration and usage details are provided through environment variables and explicit MCP connection configurations. You can connect using either the HTTP transport or a local stdio workflow. The environment variables and their purposes are listed below. When running in remote HTTP mode, include your API key for authentication. When using local development, you may omit the API key if auth is disabled on the backend.

Key environment variables and defaults you will encounter include: LEVELANG_API_BASE_URL for the backend URL, LEVELANG_API_KEY for the service key, MCP_TRANSPORT to choose the transport method, MCP_PORT for the HTTP port, and MCP_API_KEYS to enable static API key authentication for HTTP transport. Use these in your MCP client configuration blocks as shown in the example configurations.

Notes on usage patterns

- Use the HTTP transport when you want a remote, centralized MCP server reachable at a URL. Include an Authorization header when MCP_API_KEYS is configured. - Use the stdio transport for local development or when you prefer spawning a subprocess that runs the MCP server directly. In this mode, you typically run UV with a Python module that hosts the MCP server.

Available tools

Translate

Translate text through the Levelang backend with selected learner level and mood constraints.

Languages Details

Query available languages and their details from the backend.

Languages List

Query supported languages and their codes.

Context Endpoints

Tools to pull language configurations into context via MCP resources.