name: professional-translator
description: "Translate text professionally — preserving tone, register, and meaning, not word-for-word. Use when asked to translate a document, email, or content between languages, or to improve a literal/machine translation. Produces a natural, register-appropriate translation plus translator's notes on choices, untranslatable terms, and anything that needs localization rather than translation."
Professional Translator Skill
Machine translation is literal; professional translation conveys meaning, tone, and intent the way a
native speaker would say it. This skill translates with attention to register (formal vs. casual), the
audience, and idiom — and flags the places where a straight translation would mislead and a localization
choice is needed instead. (For marketing copy that must land emotionally in-culture, use
transcreation; for adapting a whole product, localization-brief.)
Required Inputs
Ask for these only if they aren't already provided:
- The text and the source → target language (incl. regional variant where it matters — e.g. Simplified vs. Traditional Chinese, LATAM vs. European Spanish).
- Register / audience — formal (legal, business), neutral, or casual; who reads it.
- Context — what it is (email, contract, UI string, marketing, instructions) — it changes the choices.
- Glossary / do-not-translate terms — brand names, product terms, anything fixed.
Output Format
Translation: [source] → [target]
Translation — the natural, register-appropriate target text. Read as if originally written in the target language, not translated into it.
Translator's notes — the choices a careful translator would flag:
- Register/tone — how formality was handled (e.g. 您 vs. 你 in Chinese, tu vs. usted, keigo in Japanese).
- Untranslatable / adapted terms — what had no direct equivalent and how it was rendered.
- Localization flags — where a literal translation would be wrong or odd: idioms, dates/units/currency, examples, cultural references — and the adaptation made (or a 🔴 flag if the user must decide).
- Kept verbatim — brand names, code, identifiers, URLs, proper nouns (unchanged).
- Ambiguities — anything in the source open to interpretation, with the assumption made.
Quality Checks
Anti-Patterns
Based On
Professional translation practice — meaning-based (not literal) translation, register matching, and the translation-vs-localization distinction.
Example Trigger Phrases
- "Translate this document."
- "Translate this email into Spanish."
- "Improve this machine translation."
- "Translate this keeping the tone."