mcpField
T
translate-site

translate-site

github.com/cloudcannonDevelopment992 installsUpdated 5d ago
Repository

About translate-site

translate-site is a multilingual site translation skill for AI coding agents, covering Rosey locale JSON and per-locale content directories. It identifies untranslated entries from source-value equality and leaves completed translations intact for incremental work.

Key features

  1. 01Translate Rosey locale JSON entries
  2. 02Translate per-locale content files
  3. 03Identify untranslated or stale entries
  4. 04Keep existing translations for incremental work
  5. 05Use keys to understand text context

Use cases

  1. 01Bulk-translate entries for a new locale
  2. 02Update stale translations after source changes
  3. 03Translate bodies and metadata in locale directories

Install

Copy the command and run it in a terminal.

$npx skills add https://github.com/cloudcannon/agent-skills --skill translate-site

After installing, type/translate-sitein Claude Code to confirm it loaded.

Bundled files

References and scripts installed along with SKILL.md.

content-directories.md
locale-files.md
scripts/
frontmatter.mjs
merge-content-translation.mjs
merge-translation.mjs
prepare-content-translation.mjs
prepare-translation.mjs
README.md

Details

From the project README

Translate a Multilingual Site with AI

Two translation systems can run on the same site, and this skill covers both:

PartWhat it holdsSection
Rosey locale JSON (rosey/locales/{code}.json)Shared UI and any page text tagged with data-rosey — nav, footer, headings, buttons, breadcrumbs, and (for non-split pages) whole page bodiesPart 1
Content collection files (blog_fr/, content_fr/ …)Per-locale body content + frontmatter (title, description, alt text, page-builder text) that the SSG renders natively per localePart 2

Most Rosey sites only need Part 1. Only reach for Part 2 if the project actually has per-locale content directories: split-by-directory (the make-site-multilingual skill, Phase 8), or Hugo languages (its Hugo setups A and B). On a Rosey site, a split-by-directory page needs both: its body comes from a content collection file (Part 2), its shared UI from the locale JSON (Part 1). A Hugo site without Rosey (setup A) needs only Part 2.

This skill is written for AI coding agents, but the process works with any AI tool that reads and writes JSON.

When to use

  • Locale files have untranslated entries (value still equals original, and the entry wasn't deliberately kept)
  • A source string changed and its translations are now stale
  • A new locale needs bulk-translating from scratch
  • Per-locale content directories (blog_fr/, blog_de/) need their bodies and frontmatter translated

When not to use

  • The site has nothing to translate into yet — no rosey/locales/, and no per-locale content directories. Set it up with make-site-multilingual first.
  • Adding a new locale to the build — that is a pipeline and config change, also make-site-multilingual
  • Translations come from a human team or an external service — this skill only covers filling the files with AI

Contents

FileCovers
SKILL.md (this file)Which part applies, and why the file format suits AI translation
locale-files.mdPart 1, most sites need only this — rosey/locales/{code}.json
content-directories.mdPart 2 — per-locale content directories, only if the site is split-by-directory
scripts/README.mdThe prepare/merge scripts both parts drive

Why Rosey files are ideal for AI

Rosey locale files are flat JSON with a predictable three-field structure per entry. That gives an agent three properties that make translation efficient and safe:

  1. Incremental — new entries from write-locales have value set to the source original. Comparing value to original instantly identifies what's untranslated; already-translated entries are left untouched. No diffing, no external state, no tracking database.
  2. Deterministic / idempotent — running the same pass twice produces the same output. No re-translation of existing work, reviewable git diff.
  3. Context-rich — keys encode where text appears (nav:about, index:hero:title, blog:recent-posts), which disambiguates short strings ("More", "Back", "Home") without a screenshot.

The data format is the state management: read a JSON file, find entries where value === original, translate them, write the file. The one exception is an entry deliberately left the same as its original (a brand name); merge-translation.mjs records those in rosey/translate-site-keep.json so they aren't offered again.

Learnings and Gotchas

This section is a living document. When you discover new patterns, issues, or improvements while translating, ask the user before appending them here. See the repo README's Key conventions → Living documents.

Comments

Markdown supported