Douyin Prohibited Word Detection

Problem

Douyin voice-over scripts, subtitles, product listings, and livestream scripts often get throttled or removed because of one or two prohibited terms. Manual review is slow, and replacing words without checking context can make copy awkward. This skill sends text, documents, screenshots, or page links through a compliance check and returns risk points, replacement suggestions, and a usable optimized version.

How It Works

• Input modes: supports mutually exclusive content, file, and url entry points for pasted text, TXT/DOC/DOCX uploads, image text extraction, or web page scraping.
• Web and documents: web pages use Playwright for dynamic rendering with a static fallback; documents are extracted before detection.
• Detection output: returns fixed sections for results, suggestions, and optimized copy, with replacements emphasized for easier review.
• Long text handling: content over 3000 characters pauses for user choice; over 10000 characters is recommended to be split manually.
• Stability: includes English false-positive filtering, network retries, and file delivery to reduce noisy flags.

Boundaries

It is scoped to Douyin copy compliance, does not analyze visual image content, and does not accept raw PDF uploads; a REDFOX_API_KEY is required. Final replacements should still be reviewed against brand tone and platform rules.