MCP Server

polyglot-gpu includes a built-in MCP (Model Context Protocol) server powered by FastMCP. This lets Claude Code, Claude Desktop, and other MCP clients translate text directly.

Setup for Claude Code

Add to your MCP config:

{
  "mcpServers": {
    "polyglot-gpu": {
      "command": "polyglot-gpu"
    }
  }
}

Or run the server directly:

python -m pypolyglot

The server uses stdio transport by default.

Available tools

Tool	Description
translate_text	Translate text between any of 57 languages
translate_md	Translate markdown while preserving structure
translate_all_langs	Translate into multiple languages at once
list_languages	List all 57 supported languages with codes
check_status	Check Ollama availability and installed models

translate_text

Translates plain text between two languages.

Parameters:

• text (string, required) — Text to translate
• from_lang (string, required) — Source language code (e.g., “en”)
• to_lang (string, required) — Target language code (e.g., “ja”)
• model (string, optional) — Ollama model override
• glossary (array, optional) — Custom glossary entries

translate_md

Translates markdown content while preserving code blocks, tables, HTML, and structure.

Parameters:

• markdown (string, required) — Markdown content to translate
• from_lang (string, required) — Source language code
• to_lang (string, required) — Target language code
• model (string, optional) — Ollama model override

translate_all_langs

Translates markdown into multiple languages concurrently. Injects language nav bars into each translation.

Parameters:

• markdown (string, required) — Markdown content
• from_lang (string, optional, default: “en”) — Source language
• languages (array, optional) — Target language codes (default: 7 languages)
• model (string, optional) — Ollama model override
• concurrency (number, optional, default: 2) — Max concurrent translations (1-3)

list_languages

Returns all 57 supported language codes and names. No parameters.

check_status

Reports Ollama availability, installed TranslateGemma models, and GPU readiness. No parameters.