FAQ

What is yaku?

yaku (訳, Japanese for “translation”) is a CLI tool that translates text, files, and developer artifacts from your terminal. It reads from stdin, files, or arguments and writes translated text to stdout — like jq or sed, but for natural language.

How is yaku different from other translation tools?

See Why yaku? for a detailed comparison with Google Translate, general-purpose AI CLIs (llm, sgpt, aichat), and Translate Shell.

Is yaku free?

The CLI is open source (BSD 3-Clause license) and free to use with your own API key. The hosted service at api.yakulang.com works with no setup and includes a free tier. See Hosted Service & Plans for tier details and quota limits.

Which LLMs does yaku support?

Four backends:

Backend	Default model	How to use
Hosted (`api.yakulang.com`)	Server-side	Default — works with no setup
Google Gemini	`gemini-2.5-flash`	`yaku config set backend gemini`
OpenAI	`gpt-4o-mini`	`yaku config set backend openai`
Anthropic	`claude-haiku-4-5-20251001`	`yaku config set backend anthropic`

The OpenAI backend also works with any OpenAI-compatible API (Groq, Together.ai, DeepSeek, Ollama). See Backends.

Which languages does yaku support?

Any language your chosen LLM backend can translate. yaku does not maintain a hardcoded language list — it passes the language code directly to the model. See Languages.

Does yaku store my text?

Local backends (Gemini, OpenAI, Anthropic): yaku sends your text directly to the LLM API and returns the result. Nothing is stored by yaku. Review your LLM provider’s privacy policy for their data handling.

Hosted backend: your text is sent to api.yakulang.com for translation. The hosted service does not persist your text after the response is returned.

Can I use yaku offline?

Yes, with a local model via the OpenAI-compatible backend. For example, with Ollama:

yaku --backend openai \
  --api-base http://localhost:11434/v1 \
  --model llama3 \
  --to en "Bonjour"

See Backends — OpenAI-compatible providers.

How do I ensure consistent terminology?

Use a glossary file. Create .yaku-glossary.yaml in your project root:

zh-TW:
  Kubernetes: ~     # keep in English
  container: 容器   # always translate to this

yaku auto-loads it and injects the terms into every translation prompt.

Can I use my own LLM prompt?

Yes, with local backends. The --prompt flag replaces yaku’s built-in system prompt entirely:

yaku --to en --prompt my-prompt.txt -f docs.ja.md

Custom prompts only work with local backends (Gemini, OpenAI, Anthropic). The hosted backend uses server-side prompts and ignores --prompt. See Custom Prompts.

How do I report a bug?

Open an issue with your yaku version (yaku version), the command you ran, and the error message.