- Home
- Skills
- Translation & Localization
- i18n String Extraction and Setup for React Next Vue Flutter
Works with the AI Agents you already use
i18n String Extraction and Setup for React Next Vue Flutter
Find hard-coded strings, extract them to ICU message catalogs, wire the right i18n library and add a CI gate. Verdict LOCALIZED / PARTIAL / HARD-CODED.
$5
i18n String Extraction and Setup for React Next Vue Flutter
Example session with this skill installed
Our React + Vite dashboard is English-only. Set up i18n with react-i18next for English, German and Arabic, extract the hard-coded strings in src/features/billing and src/components, and add a CI check so new hard-coded text fails the build.
- Read your context and instructions
- Compiled the i18n string extraction
i18n Report: acme-dashboard — PARTIAL
Stack: React 18 + Vite ·
Library: i18next + react-i18next, i18next-cli (versions checked on npm) ·
Locales: en → de, ar, plus pseudo en-XA
Inventory
| Area | UI strings found | Extracted | Remaining | Notes |
|---|---|---|---|---|
| features/billing | 64 | 58 | 6 | 6 in PDF invoice template (server-rendered) |
| components | 41 | 41 | 0 | |
| Total | 105 | 99 | 6 | 23 technical strings left as-is |
Changes made
- Added
src/i18n.ts(i18next init,fallbackLng: { 'de-AT': ['de'], default: ['en'] }, lazy-loaded namespaces) - Added
public/locales/{en,de,ar,en-XA}/billing.json,common.json - Added
i18next.config.ts,.github/workflows/i18n.yml <html lang dir>now set on language change
Before → after
// before
<p>{count} invoice{count === 1 ? '' : 's'} due on {d.toLocaleDateString()}</p>
// after
<p>{t('billing.dueSummary', { count, date: d })}</p>
"dueSummary_one": "{{count}} invoice due on {{date, datetime}}",
"dueSummary_other": "{{count}} invoices due on {{date, datetime}}"
Prices: '$' + amount.toFixed(2) → new Intl.NumberFormat(i18n.language, { style: 'currency', currency }).format(amount)
RTL
- 14 physical CSS properties converted to logical (
margin-inline-start,inset-inline-end) - Back/forward chevrons mirrored under
[dir=rtl]; logos and charts untouched
CI gate
- run: npx i18next-cli lint
- run: npx i18next-cli extract --ci --dry-run
- run: npx i18next-cli status en
Fails on hard-coded JSX text, keys missing from en, or catalogs out of date.
Draft translations — NEEDS HUMAN REVIEW
None created (not requested). de and ar catalogs contain empty values.
Remaining work
server/templates/invoice.hbs— 6 strings, needs server-side i18next instance- Pseudo-locale run:
PlanCardtitle clips at +35% length; allow wrapping
Connects securely to your tools. The creator never sees your data.
About this skill
Adding translations to an app that was built in one language is slow and easy to get wrong. Strings hide in JSX, attributes and template literals, plurals get built with + 's', dates are formatted by hand, and new hard-coded text creeps back in the week after the migration.
This skill gives your coding agent a fixed, checkable process for internationalizing a codebase:
- Detects the stack and picks the matching library: react-i18next, FormatJS, Lingui, next-intl, vue-i18n, Angular localize, Paraglide JS, Django gettext, Rails I18n, Flutter ARB, iOS String Catalogs or Android strings.xml
- Inventories hard-coded user-facing strings and separates them from technical strings that should stay
- Extracts strings with semantic keys, whole-sentence interpolation and ICU plural/select syntax, with translator descriptions
- Replaces hand-built dates, numbers and currency with Intl APIs
- Sets up locale negotiation, fallback chains, the lang and dir attributes, and logical CSS for RTL
- Adds a pseudo-locale to expose missed strings and layout overflow
- Adds a CI job that fails on new hard-coded strings, missing keys or invalid ICU syntax
- Ends with a report: inventory table, before/after diffs, remaining work and a LOCALIZED, PARTIAL or HARD-CODED verdict
It is for frontend, full-stack and mobile developers preparing a first translation hand-off, or teams auditing how localizable an app is.
Honesty note: the skill does not machine-translate silently. If you ask for draft translations they are flagged for human review. It tells the agent to check library versions against official docs and to mark anything it could not verify. Commands in the bundled reference were checked against official docs in October 2026 and may change in later releases.
It does not produce final translations or replace a translator or TMS. Static detection misses strings built at runtime or stored in databases and CMSs, and server-rendered templates may need separate work. Library commands change between major versions, so results depend on the agent checking current docs; without web access it falls back to the bundled October 2026 reference.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before any payment.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Money-back guarantee
Doesn't do what its page promises? You get your money back.
Trust & safety
Security scan
Security scan passed 29 days ago
Requires