I wanted to tell my girlfriend something real in her first language — not phrasebook-correct, but actually natural. Google Translate gave me one sentence, no register, no note on whether it sounded stiff or weirdly intimate. I had no way to know what I was actually sending.

That's the failure mode of literal translation tools: they ship you an answer without teaching you anything about it. You get through the task, you learn nothing, and if the register is wrong you don't find out until you've already said it.

So I built konid. It returns three options for anything you want to say — ordered casual to formal — with the register explained and cultural context comparing the choices. For the Mandarin and Japanese options especially, it also plays audio pronunciation through your speakers directly, using node-edge-tts with no external API key. Hearing the tone contour on a Mandarin phrase is the difference between storing the written form and actually remembering how the word moves — your brain encodes the sound pattern differently than it encodes a romanization.

The workflow for someone translating work emails in a language they're learning: you write the sentence you want, konid returns the three variants with notes on when each one is appropriate, and you pick with context instead of guessing. Over time that compounds — you start recognizing register from exposure instead of having to ask.

Installs as an MCP server in one command:

claude mcp add konid-ai -- npx -y konid-ai

Works with Claude Code, Cursor, VS Code Copilot, Windsurf, Zed, JetBrains, and Claude Cowork. Also available as a ChatGPT app via Developer mode at the endpoint https://konid.fly.dev/mcp. Supports 13+ languages: Mandarin, Japanese, Korean, Spanish, French, German, Portuguese, Italian, Russian, Arabic, Hindi, and more.

MIT licensed. https://github.com/robertnowell/konid-language-learning