I was learning Farsi and wanted to tell my girlfriend I'd missed her that day. Google Translate gave me one answer, no context, no sense of whether it was warm or clinical or something you'd say to a coworker. I had no way to know if what I sent would land the way I meant it.
That's the gap konid fills. You give it a phrase — anything from "I missed you today" to a work email sign-off you need to get right — and it returns three options ordered casual to formal, with the register explained for each one and a cultural note on what the choice signals.
For "I missed you today" in Japanese, that distinction matters a lot. The casual option might drop the subject entirely, which reads as natural between people who are close. The formal option preserves it, which can feel stiff with a partner but appropriate with a supervisor. The middle option is what you'd send to a friend you've been dating for two years. konid names that instead of leaving you to guess.
Audio pronunciation plays directly through your speakers via node-edge-tts — no external API key, no copying phonetics into a separate tab.
I use it daily for work emails in a language I'm still learning. Before, I'd get the translation done and learn nothing. Now I read the nuance comparison between the three options and actually understand why one is better for the context. That's the habit that builds the language.
It works as an MCP server — Claude Code, Cursor, VS Code Copilot, Windsurf, Zed, JetBrains, Claude Cowork — and as a ChatGPT app via Developer mode pointing at https://konid.fly.dev/mcp. Covers 13+ languages: Mandarin, Japanese, Korean, Spanish, French, German, Portuguese, Italian, Russian, Arabic, Hindi, and more.
claude mcp add konid-ai -- npx -y konid-ai
MIT licensed. https://github.com/robertnowell/konid-language-learning
Top comments (0)