Crowdin

Crowdin is a translation platform: you upload the default locale's files, translators work in Crowdin, and you download the other locales back. The module reads plain JSON files, so the Crowdin CLI only needs a crowdin.yml that maps locales/ and locales/pages/ correctly. No module setting changes.

Configure crowdin.yml

# crowdin.yml
project_id_env: CROWDIN_PROJECT_ID
api_token_env: CROWDIN_PERSONAL_TOKEN
base_path: ./locales
preserve_hierarchy: true

files:
  - source: /en.json
    translation: /%two_letters_code%.json
  - source: /pages/**/en.json
    translation: /pages/**/%two_letters_code%.json

• en is your default locale; its files are the sources.
• project_id_env / api_token_env read the ID and token from environment variables, so no secret is committed.
• preserve_hierarchy: true keeps pages/<page>/ folders as they are.
• The translated file name must equal the locale code in nuxt.config.ts. %two_letters_code% fits codes like de; for codes like pt-BR use %locale%, or map Crowdin's codes to yours with languages_mapping.

Upload and download

npm install -g @crowdin/cli

export CROWDIN_PROJECT_ID=123456
export CROWDIN_PERSONAL_TOKEN=...   # from Crowdin account settings

crowdin upload sources   # push locales/en.json and locales/pages/**/en.json
crowdin download         # write locales/de.json, locales/pages/**/de.json, ...

Run both in CI to keep files current. Downloaded files land in your repository like hand-edited ones; the next build or dev reload picks them up.

Before uploading

Clean the sources first so translators do not work on dead keys: i18n-micro clean removes unused keys, i18n-micro check reports gaps after a download. The CLI has no Crowdin integration of its own: this page is the supported way.

Related