I translate work emails every day in a language I'm still learning. For a long time I used Google Translate, which gave me one answer — usually technically correct, often weirdly stiff — and left me with no way to know whether I'd written like a colleague or like a terms-of-service document.
So I built konid: an MCP server that returns three translations per query, ordered casual to formal, with the register explained and cultural context comparing the options. It plays audio pronunciation through your speakers via node-edge-tts, no API key required.
The angle I want to walk through here is the MCP setup, because it's where the tool earns its keep across different workflows.
One install, four clients
konid runs as an MCP server. Add it to Claude Code once:
claude mcp add konid-ai -- npx -y konid-ai
That same server also works in Cursor, VS Code Copilot, Windsurf, Zed, JetBrains, and Claude Cowork — paste the config in once per client and you're done. For ChatGPT via Developer mode, point it at https://konid.fly.dev/mcp as the endpoint.
The reason this matters: I was already living in Claude Code for code work and Claude Cowork for longer writing. I didn't want to context-switch to a browser tab every time I needed to phrase something in French or Korean. Having the same translation brain accessible in the same interface I'm already using keeps the feedback loop tight.
What a response looks like
Ask konid how to say 'I'm sorry for the late reply' in Japanese and it returns three options. The first is the kind you'd use with a friend you've known for years. The second sits in the polite-neutral register you'd use with a colleague. The third is keigo — the formal register that acknowledges hierarchy. Each one comes with a note on when and why you'd choose it, and audio pronunciation plays directly through your speakers.
Literal translation tools give you one answer because they're optimized for throughput. The one answer is fine if you want to get through a task. It's not fine if the task is making someone feel like you actually thought about how to talk to them.
Language coverage
Currently 13+ languages: Mandarin, Japanese, Korean, Spanish, French, German, Portuguese, Italian, Russian, Arabic, Hindi, Farsi, and more. MIT licensed.
Top comments (0)