A coding agent localizes a feature: confidence scores, review and the record, step by step

A real run, not a diagram: Claude Code adds a snooze feature to a React app, reports eleven keys to Locize over MCP, and gets them back in six languages within seconds, each with a confidence score. What the scores caught, what they missed, what the agent fixed by itself, what the review still changed, and what the provenance export records about each string.