Review a UI translation with AI
Compare source strings, translated copy and the screen where people read them. Spot terminology drift, lost placeholders and awkward line breaks.
Review a UI translation with AI when a sentence is grammatically sound but seems to promise the wrong action. The translation-quality-review skill checks meaning against the source and examines the string beside the actual heading or control. A button intended to save a selection should not read like the final step in placing an order.
The same review can catch a missing count placeholder or an awkward line break that only appears at a narrow width. Give the agent a glossary if your product has one. A fluent reviewer who understands the subject still makes the final judgement about specialised terms and tone.
What it does
- Compare the promise of a translated button with its actual action.
- Check whether formal and informal address vary across screens.
- Find inconsistent translations of a product term.
- Inspect clipped labels and line breaks in the interface.
- Verify that variable placeholders and required plural forms survive.
Install and use
- Install the skill in your agent and supply the source, target text and screen context.
- Ask, “Review this file-selection message. Can users tell what the count means, and does the tone match the heading?”
- Try the proposed wording on the screen and ask a native speaker familiar with the product to check it.
FAQ
Source
Made by Amey Thakur and Archit Konde. Original skill: translation-quality-review, published on GitHub under the MIT licence.
View on GitHub