Close the translation gaps — drafts created, linked, reviewed

The job

Multilingual sites drift: pages published in one language only, translations two versions behind, a brand term translated three different ways. NibWP reads translation status through WPML or TranslatePress — or audits a Weglot setup — and closes the gaps as reviewable drafts.

Human translators keep the last word; the agent does the finding, drafting and linking.

The prompts, exactly as you'd type them

Plain English is the interface. These are real prompts for this job — copy them, swap in your own names, and adjust in conversation.

The gap report

Which pages and guides have no French translation, and which have a translation older than the source? Table by traffic.

The backfill

Create German drafts for every guide that has none, translated from the English source with our glossary, linked to the original. Drafts only.

Consistency

Everywhere 'Checkout' was translated in French, use 'Paiement'. Show me each occurrence first.

The process — what actually happens

Every write follows the same shape: the agent reads before it writes, shows you a plan, and nothing lands without your approval. The audit log records every call.

1. Translation status is read from the plugin — per language, per post type.
2. Drafts are created linked to their originals, so WPML/TranslatePress show them correctly.
3. Your glossary rides along in the prompt or a skill file; terminology stays consistent.
4. On Weglot, the audit names each misconfiguration (hreflang, exclusions) with its fix.
5. Batches are shown string-by-string or post-by-post before saving.

The tools this uses

Runs on the WPML, TranslatePress and Weglot integrations:

• nibwp/wpml-* — status, create + link translations
• nibwp/translatepress-* — strings, languages, bulk updates
• nibwp/weglot-audit — misconfigurations, ranked, with fixes

The outcome

The secondary languages stop being second-class: gaps close in review sessions instead of translation sprints.

Related solutions

See also SEO at scale and content automation.