All products

Write translator context in Crowdin with AI

Explain where an ambiguous string appears and what it does. Prepare context in a local export, preview the proposed changes and upload the checked entries.

To write translator context in Crowdin with AI, give context-extraction the strings that are difficult to interpret in isolation. “Open” might describe an unresolved issue or tell someone to open a file. A translator who only sees that word needs the role of the control, not a longer paraphrase of the same word.

The agent starts with the source text, key and existing context. If a reliable file reference points to the usage and the meaning is still uncertain, it can inspect the nearby code. The workflow prepares the context entries in a local file and previews the upload before writing them to the project. A new coverage count helps the team see what still needs attention.

What it does

  • Identify ambiguous buttons, labels and status messages.
  • Explain the UI role of a selected string.
  • Use the string key and nearby source evidence.
  • Preview the entries that would be uploaded.
  • Compare context coverage after a successful write.

Install and use

  1. Install the skill in a compatible agent and configure the official Crowdin CLI with project access. An existing JSONL export can also be used.
  2. Ask “Write translator notes for ambiguous order statuses and buttons. Flag entries where the available evidence does not establish their role.”
  3. Check the dry-run list, upload the agreed changes and review the coverage report.

FAQ

Source

Made by Crowdin. Original skill: context-extraction, published on GitHub under the MIT licence.

View on GitHub