I wanted to tell a friend "I missed you" in Mandarin. Google Translate gave me one answer. I had no idea if it was the kind of thing you'd text someone or say at a funeral.
That gap is what I built konid to close.
The core problem with literal translation tools: they treat every sentence as a word-matching problem. A language is a register problem. "I missed you" between close friends sounds nothing like "I missed you" in a professional apology, and the difference isn't just vocabulary — it's tone contour, vowel length, formality markers that a flat text string can't convey.
So konid returns three options per query, ordered casual to formal, with the register difference explained in plain language. More importantly, it plays audio pronunciation through your speakers directly — no external API key, using node-edge-tts — because reading a Mandarin or Arabic romanization and hearing how it actually sounds are two completely different memory events. The tone contour and vowel length is what your brain needs to actually retain the phrase, not just copy-paste it.
Setup for Claude Code:
claude mcp add konid-ai -- npx -y konid-ai
Also works in Cursor, VS Code Copilot, Windsurf, Zed, JetBrains, and Claude Cowork. If you're on ChatGPT, there's a Developer mode endpoint at https://konid.fly.dev/mcp.
Covers 13+ languages: Mandarin, Japanese, Korean, Spanish, French, German, Portuguese, Italian, Russian, Arabic, Hindi, and more.
MIT licensed. github.com/robertnowell/konid-language-learning
Top comments (0)