feat: add Yoruba bilingual localization testing skill

skills/yoruba-localization-testing/SKILL.md

@@ -0,0 +1,82 @@
+---
+name: yoruba-localization-testing
+description: Validate Yoruba (yo) and English bilingual application interfaces, Unicode diacritics, translations, and accessible localization tests.
+origin: ECC
+---
+
+# Yoruba Localization Testing
+
+## When to Activate
+
+Use when building or reviewing Yoruba (yo) interfaces, Yoruba-English language switching, localized forms, search, notifications, or accessibility features.
+
+## Core Concepts
+
+- Preserve Yoruba tone marks and underdots: ẹ, ọ, ṣ and accented vowels such as á, à, ẹ́, ọ̀.
+- Store and serve UTF-8; use Unicode normalization consistently (NFC) when comparing strings.
+- Do not strip diacritics from displayed text or silently rewrite a speaker's wording.
+- Use BCP 47 language tags `yo` and `en`. Keep locale separate from region.
+- Avoid word-for-word translations of idioms. Have fluent Yoruba speakers review wording, context, and tone.
+- Never use machine-translated strings as the only quality gate.
+- Test long text, screen readers, keyboard navigation, font glyph coverage, and mobile layouts.
+
+## Implementation Workflow
+
+1. Inventory user-facing strings and identify hard-coded text.
+2. Extract messages into locale dictionaries with stable keys.
+3. Preserve placeholders, markup, punctuation, and pluralization semantics.
+4. Implement explicit language selection and fallback to English for missing translations.
+5. Test Unicode normalization and round-trip storage through API and database.
+6. Ask a fluent Yoruba reviewer to check accuracy and naturalness before release.
+
+## Code Examples
+
+```typescript
+const messages = {
+  en: { welcome: "Welcome", search: "Search" },
+  yo: { welcome: "Ẹ káàbọ̀", search: "Wá" },
+} as const;
+
+type Locale = keyof typeof messages;
+type Key = keyof typeof messages.en;
+
+export function translate(locale: Locale, key: Key): string {
+  return messages[locale]?.[key] ?? messages.en[key];
+}
+
+export function sameNormalizedText(a: string, b: string): boolean {
+  return a.normalize("NFC") === b.normalize("NFC");
+}
+```
+
+### Test Cases
+
+```typescript
+import { strict as assert } from "node:assert";
+
+assert.equal(translate("yo", "welcome"), "Ẹ káàbọ̀");
+assert.equal(sameNormalizedText("ẹ́", "ẹ\u0301"), true);
+assert.equal("Ọ̀rọ̀".normalize("NFC"), "Ọ̀rọ̀".normalize("NFD").normalize("NFC"));
+```
+
+## Anti-Patterns
+
+- Removing accents or underdots for convenience.
+- Assuming all Yoruba users prefer Yoruba-only navigation.
+- Translating technical identifiers, code symbols, or API keys.
+- Guessing a translation for sensitive legal, medical, or financial content.
+- Claiming localization is complete without native-speaker review.
+
+## Best Practices
+
+- Keep UTF-8 source files and translation fixtures in version control.
+- Use `Intl` APIs for locale-sensitive formatting where supported; verify runtime support.
+- Allow bilingual switching without losing form state.
+- Provide a user-visible way to report translation problems.
+- Test on actual target fonts, browsers, and assistive technologies.
+
+## Related Skills
+
+- `accessibility`
+- `frontend-patterns`
+- `tdd-workflow`