Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Translate content across languages while preserving brand voice and cultural nuance.
.claude/skills/holaboss-ai-translator/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-13 | ✗→✓ | ▲ Improved | 22% | 0% |
| case-14 | ✗→✓ | ▲ Improved | 49% | 0% |
| case-19 | ✗→✓ | ▲ Improved | 40% | 0% |
| case-07 | ✓→✓ | ＝ Same ✓ | 19% | 0% |
| case-01 | ✓→✓ | ＝ Same ✓ | 32% | 0% |
Translate like a localization professional, not a dictionary. A good translation reads as if it were originally written in the target language for that audience — the idioms land, the tone matches, and the brand still sounds like itself.
Use Translator to move content between languages while keeping it natural and on-brand: marketing copy, product UI strings, social posts, support replies, documentation. The skill handles 50+ languages and is built for localization, not just literal conversion.
The difference between translation and localization is where the value is:
When the target has limits (tweet length, button labels, push notifications, UI fields), keep the translation within them. Languages expand and contract — German and Finnish often run long, CJK often runs short — so rework phrasing to fit rather than truncating mid-thought.
{name}, %s, HTML tags) exactly.Return the translated text ready to ship, preserving the original structure and any markup. If you made a localization choice worth knowing about (a swapped idiom, a chosen regional variant, a length adaptation), add a brief translator's note after the text. For glossary or terminology decisions, keep them consistent across the whole piece.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-07 | pass→pass | 11,992 | 11,165 | -7% | 1 | 1 | 0% | 1,949 | 2,321 | +19% | 0 | 0 | — |
case-01 | pass→pass | 6,318 | 11,536 | +83% | 1 | 1 | 0% | 1,043 | 1,380 | +32% | 0 | 0 | — |
case-02 | pass→pass | 5,223 | 4,712 | -10% | 1 | 1 | 0% | 877 | 1,084 | +24% | 0 | 0 | — |
case-03 | pass→pass | 6,026 | 5,733 | -5% | 1 | 1 | 0% | 902 | 1,398 | +55% | 0 | 0 | — |
case-04 | pass→pass | 13,842 | 20,167 | +46% | 1 | 1 | 0% | 2,843 | 3,575 | +26% | 0 | 0 | — |
case-05 | pass→pass | 9,773 | 7,320 | -25% | 1 | 1 | 0% | 1,577 | 1,802 | +14% | 0 | 0 | — |
case-06 | pass→pass | 7,263 | 7,420 | +2% | 1 | 1 | 0% | 1,319 | 1,688 | +28% | 0 | 0 | — |
case-08 | pass→pass | 7,502 | 8,600 | +15% | 1 | 1 | 0% | 1,325 | 1,948 | +47% | 0 | 0 | — |
case-09 | pass→pass | 8,610 | 8,178 | -5% | 1 | 1 | 0% | 1,282 | 1,729 | +35% | 0 | 0 | — |
case-10 | fail→fail | 5,937 | 7,244 | +22% | 1 | 1 | 0% | 1,177 | 1,787 | +52% | 0 | 0 | — |
case-11 | pass→pass | 9,411 | 9,109 | -3% | 1 | 1 | 0% | 1,342 | 2,006 | +49% | 0 | 0 | — |
case-12 | pass→pass | 6,197 | 4,760 | -23% | 1 | 1 | 0% | 855 | 1,096 | +28% | 0 | 0 | — |
case-13 | fail→pass | 19,442 | 26,337 | +35% | 1 | 1 | 0% | 3,834 | 4,684 | +22% | 0 | 0 | — |
case-14 | fail→pass | 5,337 | 6,590 | +23% | 1 | 1 | 0% | 1,022 | 1,526 | +49% | 0 | 0 | — |
case-15 | fail→fail | 8,843 | 7,101 | -20% | 1 | 1 | 0% | 1,242 | 1,679 | +35% | 0 | 0 | — |
case-16 | pass→pass | 9,634 | 10,140 | +5% | 1 | 1 | 0% | 1,701 | 1,885 | +11% | 0 | 0 | — |
case-17 | pass→pass | 5,665 | 9,060 | +60% | 1 | 1 | 0% | 954 | 1,728 | +81% | 0 | 0 | — |
case-18 | pass→pass | 9,062 | 6,378 | -30% | 1 | 1 | 0% | 1,583 | 1,342 | -15% | 0 | 0 | — |
case-19 | fail→pass | 7,911 | 5,908 | -25% | 1 | 1 | 0% | 991 | 1,389 | +40% | 0 | 0 | — |
case-20 | pass→pass | 7,079 | 8,081 | +14% | 1 | 1 | 0% | 1,235 | 1,597 | +29% | 0 | 0 | — |
case-21 | pass→pass | 10,865 | 10,997 | +1% | 1 | 1 | 0% | 1,830 | 2,007 | +10% | 0 | 0 | — |
case-22 | pass→pass | 3,390 | 3,023 | -11% | 1 | 1 | 0% | 605 | 1,021 | +69% | 0 | 0 | — |
case-23 | pass→pass | 1,818 | 2,291 | +26% | 1 | 1 | 0% | 338 | 789 | +133% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 23 cases were attempted. The headline lift of +13 percentage points is the difference between those two pass rates over the 23 comparable cases.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.