Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Use when reviewing a codebase for internationalisation readiness, setting up an i18n library, or preparing strings for a new locale.
.claude/skills/thedaviddias-translation-strings/SKILL.md| Model | Eval pass | Runs |
|---|---|---|
| gemini-3.6-flash | 98% | 83 |
| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-08 | ✗→✓ | ▲ Improved | 3% | 0% |
| case-13 | ✗→✓ | ▲ Improved | -19% | 0% |
| case-11 | ✓→✓ | ＝ Same ✓ | 11% | 0% |
| case-01 | ✓→✓ | ＝ Same ✓ | -31% | 0% |
| case-02 | ✓→✓ | ＝ Same ✓ | -23% | 0% |
Naive string concatenation produces untranslatable sentences because word order varies dramatically across languages. A pattern like "Found " + count + " results" cannot be translated correctly into languages where the number appears in a different position or where the noun form changes depending on the count. Bad i18n strings force translators to work around developer mistakes — or simply leave content untranslated.
Check whether translation strings use proper message formatting with placeholders rather than string concatenation.
Replace concatenated translation strings with ICU message format patterns and ensure pluralisation is handled via the library, not conditionals.
Explain why string concatenation breaks translations and how ICU MessageFormat handles variable word order, pluralisation, and gender.
Review components and utility functions for hardcoded user-visible strings, string concatenation involving translated text, and missing plural forms.
For full implementation details, code examples, and framework-specific guidance, see references/rule.md.
Rule page: https://frontendchecklist.io/en/rules/javascript/translation-strings
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-11 | pass→pass | 7,972 | 7,664 | -4% | 1 | 1 | 0% | 1,649 | 1,824 | +11% | 0 | 0 | — |
case-01 | pass→pass | 9,808 | 5,982 | -39% | 1 | 1 | 0% | 1,912 | 1,315 | -31% | 0 | 0 | — |
case-02 | pass→pass | 8,562 | 5,443 | -36% | 1 | 1 | 0% | 1,624 | 1,244 | -23% | 0 | 0 | — |
case-03 | pass→pass | 10,759 | 7,545 | -30% | 1 | 1 | 0% | 1,802 | 1,534 | -15% | 0 | 0 | — |
case-04 | pass→pass | 7,538 | 5,261 | -30% | 1 | 1 | 0% | 1,570 | 1,290 | -18% | 0 | 0 | — |
case-05 | pass→pass | 12,011 | 7,293 | -39% | 1 | 1 | 0% | 2,088 | 1,514 | -27% | 0 | 0 | — |
case-06 | pass→pass | 5,257 | 6,808 | +30% | 1 | 1 | 0% | 1,008 | 1,508 | +50% | 0 | 0 | — |
case-07 | pass→pass | 4,771 | 6,072 | +27% | 1 | 1 | 0% | 840 | 1,307 | +56% | 0 | 0 | — |
case-08 | fail→pass | 4,805 | 3,359 | -30% | 1 | 1 | 0% | 791 | 813 | +3% | 0 | 0 | — |
case-09 | pass→pass | 4,376 | 5,552 | +27% | 1 | 1 | 0% | 800 | 1,232 | +54% | 0 | 0 | — |
case-10 | pass→pass | 8,252 | 4,994 | -39% | 1 | 1 | 0% | 1,609 | 1,193 | -26% | 0 | 0 | — |
case-12 | pass→pass | 9,425 | 5,405 | -43% | 1 | 1 | 0% | 1,551 | 1,236 | -20% | 0 | 0 | — |
case-13 | fail→pass | 13,057 | 8,561 | -34% | 1 | 1 | 0% | 2,350 | 1,900 | -19% | 0 | 0 | — |
case-14 | pass→pass | 9,101 | 7,854 | -14% | 1 | 1 | 0% | 1,638 | 1,707 | +4% | 0 | 0 | — |
case-15 | pass→pass | 8,443 | 4,865 | -42% | 1 | 1 | 0% | 1,504 | 1,169 | -22% | 0 | 0 | — |
case-16 | pass→pass | 8,611 | 6,931 | -20% | 1 | 1 | 0% | 1,620 | 1,429 | -12% | 0 | 0 | — |
case-17 | pass→pass | 15,244 | 10,685 | -30% | 1 | 1 | 0% | 2,817 | 2,285 | -19% | 0 | 0 | — |
case-18 | pass→pass | 9,328 | 7,734 | -17% | 1 | 1 | 0% | 1,728 | 1,756 | +2% | 0 | 0 | — |
case-19 | pass→pass | 9,950 | 5,142 | -48% | 1 | 1 | 0% | 1,759 | 1,237 | -30% | 0 | 0 | — |
case-20 | pass→pass | 9,706 | 7,984 | -18% | 1 | 1 | 0% | 1,780 | 1,811 | +2% | 0 | 0 | — |
case-21 | pass→pass | 11,958 | 11,977 | +0% | 1 | 1 | 0% | 2,433 | 2,826 | +16% | 0 | 0 | — |
case-22 | pass→pass | 11,406 | 9,340 | -18% | 1 | 1 | 0% | 2,016 | 2,167 | +7% | 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. 22 cases were attempted. The headline lift of +9 percentage points is the difference between those two pass rates over the 22 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.