translation-creative

Layer 2-5 owner for ad copy / marketing brief / headline / tagline / catchphrase in translation-toolkit. Reads an intake-spec produced by translation-intake (Layer 1), runs an optional brand-brief intake, then runs the shared DRAFT / REFLECT / IMPROVE core loop with mode-conditional behavior: faithful mode = 4D reflection + S1 SHOULD; transcreation mode = 5D reflection (adds Effectiveness axis) + S1 MUST. Default emits a single output; --variants=N opts into N genuinely different alternatives.

Inputs

source — ad copy / headline / tagline / marketing-brief string, or a path to a short text file.
intake-spec — from translation-intake. If invoked directly without one, run intake first (--intake flag) before continuing. Intake-spec carries mode (faithful | transcreation) — the load-bearing field for this skill.
brand brief (optional but strongly recommended for transcreation) — captured per protocols/brand-brief-intake.md (archetype, tone-of-voice, do-not-say list, signature phrases, target persona, CTA style, brand-name handling). Skill-local; not part of the four shared service-interface fields.
glossary_path (optional) — repo-local project glossary (L1 tier of the 4-tier resolver). Defaults to <caller_repo>/docs/i18n/glossary-{target_locale}.md; missing → fall through to L2 bundled glossary in glossary/.

Service-interface contract (the four shared fields plus web_search) is defined in the design spec §Service Interface; no per-skill duplication here.

When to use

Routed in by using-translation-toolkit when the input is ad-shaped (short, persuasive, headline-like) or when the user explicitly names ad copy / marketing / catchphrase / tagline / brand voice.
Invoked directly by name when the user already knows creative is the right specialist.
Not for i18n format files (use translation-i18n), prose / Markdown docs (use translation-doc), or reviewing an existing translation pair (use translation-audit).