I wanted to tell my girlfriend 'I missed you today' in Farsi. Google Translate gave me a sentence. One sentence, no register, no cultural framing, no way to know if it sounded like a person or a form letter.
That's the gap literal translation tools leave: they get you through the task but teach you nothing and give you no choices. If you're learning a language or trying to say something that actually matters to someone, one decontextualized answer isn't enough.
So I built konid β an MCP server that returns three options for anything you want to say, ordered casual to formal, each with the register explained and cultural nuance notes comparing the options. It also plays audio pronunciation through your speakers via node-edge-tts, no external API key required.
For the girlfriend example, instead of one flat translation, you get:
- a casual version you'd say to someone you're close with
- a standard version that reads naturally in most contexts
- a more formal register, with a note on when that formality would actually land
The nuance comparison is the part I kept coming back to while building it. Knowing why one phrasing is warmer than another is what eventually makes the language stick.
It supports 13+ languages including Mandarin, Japanese, Korean, Spanish, French, German, Portuguese, Italian, Russian, Arabic, Hindi, and Farsi. Installs in one line for Claude Code:
claude mcp add konid-ai -- npx -y konid-ai
Also works with Cursor, VS Code Copilot, Windsurf, Zed, JetBrains, and Claude Cowork. If you're on ChatGPT, it's available as a Developer mode app via the endpoint https://konid.fly.dev/mcp.
MIT licensed. Source and full setup instructions at https://github.com/robertnowell/konid-language-learning.
Top comments (0)