Build a Crowdin glossary with AI
Find terminology worth standardising in real product content. Review proposed entries, protected names and supporting evidence before adding them to a shared glossary.
To build a Crowdin glossary with AI, give glossary-terms-extraction the project context as well as the source text. A word that is ordinary in a general interface can carry a specific technical meaning in a music course or a design tool. Knowing which product the term belongs to helps the agent distinguish a genuine glossary entry from a common word.
The workflow looks for repeated domain vocabulary, ambiguous words and named features that should retain their spelling. It compares proposals with existing entries before any additions. If the project has paired source and target material, that can support a proposed equivalent. A term with no reliable target-language evidence can still be recorded for a translator to resolve.
What it does
- Select specialist terms from source strings and documents.
- Identify names that should stay unchanged.
- Compare candidates with current glossary entries.
- Prepare a CSV shortlist for editorial review.
- Use paired project material to support target equivalents.
Install and use
- Install the skill in a compatible agent. Connect an authorised Crowdin glossary or use the official CLI and project configuration for the CSV route.
- Ask “Find terminology in our music course interface and guide. Separate feature names from musical terms and show the shortlist first.”
- Review definitions and remove ordinary words before committing the selected entries.
FAQ
Source
Made by Crowdin. Original skill: glossary-terms-extraction, published on GitHub under the MIT licence.
View on GitHub