I18n sync

Skill jajupmochi/agent-harness/templates/personal-cite-static/.claude/skills/i18n-sync

Linlin's curated AI agent harness configuration: workflow rules, skills, hooks, plugin recommendations, tooling preferences, and project templates. Loadable into any new project so a fresh /init can pick the relevant subset.

Install

npx -y skills add jajupmochi/agent-harness --skill i18n-sync

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

1 stars1 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

What its author says it does

Copied from the file, not written here

Check key parity across locales/*.json. If divergence is detected, list missing keys per locale and offer to add stubs. Use after editing any locale file.

SKILL.md

2.6 KB, as published. Nobody here has run it

/i18n-sync

Verify all locales/*.json files have identical key trees, and add stubs where they diverge.

Why

The site loads locales/<lang>.json at runtime. If zh.json is missing a key that en.json has, the user sees the English fallback (or empty text) where the Chinese should be — visible bug.

Steps

Glob locales/*.json to find all locale files.
Validate JSON for each (jq empty) — error out if any is invalid (the jq-validate-json hook should already prevent this, but verify).

Extract key tree from en.json (the canonical):

jq 'paths(scalars) | map(tostring) | join(".")' locales/en.json | sort -u > /tmp/keys-en.txt

For each non-en locale, extract its key tree and diff against the canonical:

for f in locales/*.json; do
  [ "$f" = "locales/en.json" ] && continue
  jq 'paths(scalars) | map(tostring) | join(".")' "$f" | sort -u > /tmp/keys-$(basename "$f" .json).txt
  diff /tmp/keys-en.txt /tmp/keys-$(basename "$f" .json).txt
done

Report:
- Which locales have which missing keys
- Which locales have EXTRA keys not in en.json (potential typo)
Offer to add stubs: for each missing key in a non-en locale, add a stub value (e.g., "<TODO: translate>") so the next deploy doesn't show empty text.

Notes for the agent

en.json is the canonical key tree. Other locales must match.
DON'T add a key to en.json without also adding it to other locales. Use this skill to check parity AFTER edits.
Stubs use <TODO: translate> so they're easy to grep + replace later.
If a locale has EXTRA keys not in en.json, ask the user: did they add a key only to that locale by mistake, or should en.json get the key too?

Companion

jq-validate-json hook — runs automatically after every Write/Edit on locales/*.json files; catches syntax errors immediately
.githooks/pre-commit (optional) — a stricter version of this skill that fails the commit if parity is broken; consider adding for stricter discipline

Example output

Locale parity check:
- en.json (canonical): 142 keys
- zh.json: 142 keys ✓
- de.json: 138 keys ❌ missing: hero.subtitle, footer.copyright_year, ...
- fr.json: 143 keys ⚠ extra: about.bio_v2 (not in en.json — is this intentional?)

Want me to add stubs to de.json and ask the user about fr.json's extra key?