humanizer-zh

Humanizer-zh is a Chinese writing assistance skill installed in Claude Code. Its core files are translated from blader/humanizer, practical rules reference stop-slop, and are organized based on Wikipedia's "Signs of AI writing" guidelines. The goal is not to bypass AI detectors, but to assist in editing, reviewing, and rewriting AI-generated content to make text closer to authentic human expression. The project supports three installation methods: one-click installation via npx skills add, Git cloning to ~/.claude/skills/humanizer-zh, or manual download and placement into Claude Code's skills directory. After installation, it can be invoked in conversations using /humanizer-zh, or users can directly ask Claude to use this skill to rewrite text. Applicable forms include pasting paragraphs, rewriting by scenario, and requesting processing of file content like article.md during conversations. The README provides input-output examples for three categories: marketing copy, academic abstracts, and blog articles. The skill lists 24 common AI writing traces, divided into four categories: content patterns, language and grammar patterns, style patterns, communication patterns, and filler words, with six items each. Specific issues include exaggerating significance and trends, promotional language, vague attribution, overuse of high-frequency words like "furthermore," "crucial," and "in-depth," negative parallelism, three-item lists, excessive use of dashes or bold text, emojis, sycophantic tone, knowledge cutoff date statements, excessive hedging, and vague positive conclusions. In addition to the rule list, the documentation provides a manual rewriting process: first identify AI patterns, then rewrite problematic segments while preserving core meaning, matching the intended tone, and injecting specific personality. Key principles include making assertions, varying sentence structure and rhythm, acknowledging complexity, appropriately using first person, allowing moderate imperfection, and replacing abstract generalizations with concrete details. The project also explains adaptation to the Chinese context, such as supplementing Chinese expressions and localized examples for some English patterns.