PicTech Cross-Border E-Commerce Image Translation

Problem

Cross-border product images often contain copy in only one language, and manual text removal, re-editing, or relayout slows down listing workflows. This skill sends images with embedded text to the PicTech translation service, processes them with structured parameters, and returns summary metadata such as local paths, preview links, and online edit links.

How It Works

• Input: accepts a local image path, folder, image URL, array, or comma-separated string.
• Language: defaults from Chinese to English; supported source languages include Chinese, Traditional Chinese, English, Thai, Russian, Indonesian, Malay, Portuguese, Spanish, French, and German, with target mappings also covering Korean, Italian, and others.
• Main-image protection: set mainImageProtection: true when the product body, logo, or key visual elements should be preserved.
• Credentials: the persistent config.vk value is required; users should configure it once rather than paste it per request, and it must not be exposed in logs or summaries.
• Output: the executor returns a structured object with status, counts.noText, local paths, preview links, and online edit links; backend status=300 means no translatable text and should be treated as a terminal skipped state.

Boundaries

The skill does not rewrite the original input image and does not guarantee perfect layout preservation. Unsupported formats, oversized files, or unsupported languages should be marked as individual failures while the remaining images continue processing. For LLM-facing output, summarize success, skipped, and failed counts plus file locations instead of dumping the full raw response.