Deepl

You can run a Model Context Protocol (MCP) server that uses the DeepL API to translate text, translate documents, rephrase content, and manage glossaries. This MCP server exposes translation capabilities you can access from MCP clients, enabling seamless multilingual communication and content localization.

How to use

You interact with the MCP server through an MCP client. Start the server using one of the provided runtime methods, then configure your client to connect to the local MCP endpoint. Your client will send translation requests, glossary lookups, or rephrasing tasks, and the server will return translated text, translated documents, or glossary-aware results using the DeepL API.

How to install

Prerequisites: make sure you have Node.js installed on your machine. You will also need a DeepL API key to enable translation features.

Additional sections

Configuration steps center on providing your DeepL API key and choosing how you run the MCP server. The server supports multiple runtime methods to start, depending on whether you prefer using a package runner or a direct node invocation.

Configuration

DeepL API Key You must obtain a DeepL API key to use translation features. Sign up for a DeepL API plan to access translation capabilities, including a free tier where available. After you have an API key, supply it to the MCP server so requests to the DeepL API are authenticated.

Security considerations

Protect your DeepL API key and limit exposure of the key in client configurations. Store sensitive values in secure locations and avoid committing keys to version control. When deploying, consider restricting access to the MCP endpoint and auditing usage.

Examples and notes

You can run the MCP server locally to experiment with translations and glossary features. The server supports automatic language detection, formality controls, and glossary usage to ensure terminology consistency across translations.

Troubleshooting

If you encounter issues with the DeepL API, verify that your API key is correct, ensure you have sufficient quota, and confirm that the language codes you use are supported by the DeepL API.

End-user guidance

Use the MCP client's translation-related tools to translate text, rephrase content, or translate documents. You can specify source and target languages, apply formality preferences, and use glossaries to enforce consistent terminology.

Tools and commands to run the MCP server

{
  "mcpServers": {
    "deepl": {
      "command": "npx",
      "args": ["deepl-mcp-server"],
      "env": {
        "DEEPL_API_KEY": "{YOUR_API_KEY}"
      }
    }
  }
}

Alternate runtime option using a local copy

{
  "mcpServers": {
    "deepl": {
      "command": "node",
      "args": ["/{ABSOLUTE_PATH_TO_SERVER}/deepl-mcp-server/src/index.mjs"],
      "env": {
        "DEEPL_API_KEY": "{YOUR_API_KEY}"
      }
    }
  }
}

Additional runtime option using a direct path with npx

npx /{ABSOLUTE_PATH_TO_SERVER}/deepl-mcp-server

Available tools

translate-text

Translate text between languages with optional automatic source detection, target language, and formality control. Supports optional glossary usage.

translate-document

Translate documents in formats such as PDF, DOCX, PPTX, XLSX, HTML, and TXT. Returns translation status and output path.

rephrase-text

Rephrase text with configurable writing style and tone to produce variations of the original content.

get-source-languages

Retrieve the full list of supported source languages with names and ISO-639 codes.

get-target-languages

Retrieve the full list of supported target languages with names and ISO-639 codes.

list-glossaries

List all glossaries in your DeepL account along with metadata including IDs and dictionaries.

get-glossary-info

Fetch metadata for a specific glossary by its ID.

get-glossary-dictionary-entries

Retrieve all dictionary entries for a glossary and a specified language pair.