Research and generate SEO-optimized localized pages for new languages. Keywords are validated through multi-dimensional analysis including Autocomplete, Top 30 competitor analysis, and market-specific search engines (Baidu for China, Naver for Korea, Yandex for Russia, Google for others). Use when (1) creating new localized content for additional languages, (2) researching keywords for target locales, or (3) validating keyword mappings.
Creates SEO-optimized localized pages through researched keywords, not translations.
Every keyword must be researched and validated, not translated.
Same concept, different expressions per language:
Why: Direct translation produces low-volume keywords that users don't actually search.
# List existing keyword files
ls .claude/skills/i18n-seo-localizer/data/
# Expected files for 14 locales
keywords-base.json (en), keywords-zh.json, keywords-zh-TW.json,
keywords-ja.json, keywords-ko.json, keywords-de.json, keywords-es.json,
keywords-fr.json, keywords-pt.json, keywords-it.json, keywords-ar.json,
keywords-id.json, keywords-ru.json, keywords-hi.json
If keyword file exists: Read and apply keywords directly
If keyword file missing: Run keyword research workflow (see Phase 1 below)
MANDATORY before generating any localized content.
High-level steps:
See keyword-research-workflow.md for complete workflow.
CRITICAL: Use the search engine your target market actually uses.
See search-engines-config.md for complete mapping.
Required fields:
⚠️ Extended fields are REQUIRED to prevent keyword stuffing.
See keyword-template.md for complete template.
After keywords validated:
pnpm build| Code | Language | Country Code | Priority | Key Insight |
|---|---|---|---|---|
| zh | Simplified Chinese | CN | T1 | 口语化简洁 |
| zh-TW | Traditional Chinese | TW | T2 | Taiwan/HK market |
| ja | Japanese | JP | T1 | イラスト vs 画像 |
| ko | Korean | KR | T1 | Tech-savvy |
| de | German | DE | T1 | KI not AI |
| es | Spanish | ES | T2 | LATAM + Spain |
| pt | Portuguese | BR | T2 | Brazil focus |
| fr | French | FR | T2 | Growing |
| it | Italian | IT | T3 | European |
| ar | Arabic | SA | T2 | RTL + مجاني |
| id | Indonesian | ID | T2 | Gratis critical |
| ru | Russian | RU | T2 | Нейросеть term |
| hi | Hindi | IN | T3 | Hinglish mix |
See 01-language-profiles.md for detailed profiles.
When generating pages:
Automatically uses keyword mappings when localizing existing English pages.
Always use explicit locale in getTranslations() calls.
See .claude/skills/shipany-page-builder/references/04-i18n-best-practices.md
.claude/skills/i18n-seo-localizer/
├── SKILL.md # This file
├── references/
│ ├── keyword-research-workflow.md # 5-metric validation workflow
│ ├── search-engines-config.md # Market-specific search engines
│ ├── keyword-template.md # JSON template and fields
│ ├── 01-language-profiles.md # Per-language guidelines
│ └── 02-seo-guidelines.md # SEO requirements
└── data/
├── keywords-base.json # English base
├── keywords-zh.json # Simplified Chinese
├── keywords-zh-TW.json # Traditional Chinese
└── [12 more locale files]
1. Check references/search-engines-config.md for correct search engine
2. Read references/keyword-research-workflow.md
3. Generate 3-5 candidates
4. Multi-dimensional validation (100-point system):
a. Search results volume (relative comparison)
b. Top 30 competitor analysis (three-tier weighting)
c. Autocomplete ranking (multiple prefix tests)
d. Related searches check
e. Language naturalness assessment
5. Calculate total score and tier consistency bonus
6. Document in keywords-{locale}.json with:
- searchEngine (e.g., "baidu", "naver")
- searchEngineMarket (e.g., "中国大陆")
- validationScores object
1. Check search-engines-config.md for target market
2. Create data/keywords-{locale}.json from template
3. For each keyword in keywords-base.json:
a. Run keyword research workflow
b. Use correct search engine for validation
c. Perform 5-metric validation
d. Document with all required fields including:
- searchEngine
- searchEngineMarket
- validationScores
4. Add language profile to references/
5. Test with sample content
6. Register in project config
Keyword Research:
Language Guidance:
Read these documents AS NEEDED during workflow execution.
User request → Check keyword file exists
↓ ↓
No Yes
↓ ↓
Research workflow Read & apply keywords
↓ ↓
Select search Generate content
engine (Baidu/ ↓
Naver/Google) Build & validate
↓
Multi-dimensional
validation (5 metrics)
↓
Document with scores