AI Agent Hub
Back to plugins
🤖

avoid-ai-writing-cn

Model Inference Updated 2026.08.29

Run the following command in DeepSeek Harness:

dsh plugin install hellogit2021/avoid-ai-writing-cn

Paste the following prompt into your AI chat to install this plugin:

Install this plugin in DeepSeek Harness by running dsh plugin --profile web add github:hellogit2021/avoid-ai-writing-cn; the source is hosted at https://github.com/hellogit2021/avoid-ai-writing-cn

About this plugin

AI-generated Chinese often carries a recognizable machine fingerprint: overused phrases like "let us delve deeper", rigid "not A, but B" templates, and vague wrap-up sentences that make the text read like a model hitting a word count rather than a person making a point. This DSH plugin exists to strip exactly that feel away.\n\nThe workflow is two sentences long. Paste your text, say "remove the AI flavor", and get a rewritten draft instantly with no fancy analysis UI. Once the draft is final, simply say "well written" and any newly discovered AI-isms and template patterns are appended to your personal avoidance list, so the next pass skips them automatically. The base rule set is maintained by the Zhihu "De-AI Writing" community and keeps evolving, while your personally accumulated patterns survive every plugin upgrade, making the output more and more aligned with your own voice over time.\n\nIt is built for anyone who relies on LLMs for everyday Chinese writing — blog posts, social media, emails, meeting notes — and wants the result to sound like a human wrote it, not a model filling in the blanks.

Use Cases

  • One-click removal of AI boilerplate and template phrases before publishing a blog draft
  • Quickly replacing vague closings with concrete expressions before finalizing a social media post
  • Making emails or meeting notes sound like a real person before sending

Best For

  • Content creators who rely on LLMs for daily Chinese writing
  • Non-native writers who want AI output to read like a human wrote it
  • Professionals who don't want colleagues to detect machine-generated text