translation

translation guides the workflow, terminology, style, and quality of multilingual content. It solves the problem of producing consistent, on-brand translations across languages by structuring the process and helping decide between human and machine translation. It complements the localization-strategy skill: this skill produces the translated content, while i18n implementation, hreflang, and URL structure are handled there. As an initial step it checks for project context in .claude/project-context.md or .cursor/project-context.md to pick up brand voice, target markets, and product terminology, then identifies content type, target locales, and existing assets like glossaries, style guides, and translation memory.

It centers on a Brief to Translate to Review workflow, where the brief captures context, audience, tone, glossary reference, style guide, and constraints such as UI character limits and SEO keywords; translation is a first pass using glossary and TM; and review is a native-speaker check for consistency, brand voice, and SEO. It provides content-type guidance (product UI is glossary-critical and avoids machine translation, marketing and landing pages use human translation, blogs require re-researching keywords in the target language rather than translating keyword lists, technical docs can use MT plus post-edit at high volume). It details glossary structure (source term, target term, context, do-not-translate), market-specific terminology examples (for instance AIGC vs Generative AI, KOL vs Influencer for Chinese), style-guide elements (voice, tone, register such as formal vs informal forms, punctuation, formatting, forbidden terms), and per-language considerations for Chinese, German, Japanese, and Arabic. It also covers when human versus machine translation is appropriate, translation memory and TMS usage, and a quality checklist plus SEO practices.

It targets marketers, localization managers, and content teams translating or localizing content, managing terminology, or optimizing translation quality. Its output format includes a translator brief, glossary additions or updates, style-guide notes, a human-vs-MT recommendation, and a quality checklist. It repeatedly cross-references related skills including localization-strategy, content-strategy, keyword-research, title-tag, and meta-description, and warns against raw MT for product and marketing content, MT without post-editing for customer-facing content, and publishing many thin translated pages at once.