I built konid after noticing a pattern: I could recognize a word in French and still mispronounce it for months. Reading a translation doesn't wire the sound. Hearing the vowel length and stress pattern does — and that gap between reading and hearing is where most translation tools leave you.
konid is an MCP server that returns 3 options per query, ordered casual to formal, with the register explained and cultural context comparing the options. Audio plays directly through your speakers via node-edge-tts — no external API key, no copy-paste into a pronunciation site.
The audio angle matters because tone contour is meaning in some languages. In Mandarin, the same syllable is a different word depending on pitch. In Arabic, vowel length changes semantics. A text-only translation hides that completely.
Here's how to add it to Claude Code:
claude mcp add konid-ai -- npx -y konid-ai
Also works in Cursor, Windsurf, VS Code Copilot, Zed, JetBrains, and as a ChatGPT app via https://konid.fly.dev/mcp.
The use case I built it for: I was writing a message to my girlfriend in her first language and needed the difference between warm-familiar and awkward-formal. Phrasebooks don't surface that. One translation answer doesn't surface that. You need the options side by side with the explanation of what each one signals to a native speaker.
13+ languages including Mandarin, Japanese, Korean, Spanish, French, Arabic, and Hindi.
Top comments (0)