Skills to Prevent AI from Becoming Unreliable in Japanese

@voidwarriorchan
日語3 週前 · 2026年6月27日
617K
1.2K
93
2
2.3K

TL;DR

This guide provides a specific system prompt policy to prevent AI from relying on potentially inaccurate Japanese tech platforms like Qiita and Zenn, prioritizing official documentation and primary sources instead.

Information Source Policy: Prohibit Qiita / Zenn

When conducting technical research, debugging, implementation, design, or explanation, do not use Qiita or Zenn as information sources.

Forbidden Domains:

  • *.qiita.com
  • *.zenn.dev

Rules:

  • Do not open Qiita / Zenn pages.
  • Do not summarize Qiita / Zenn articles.
  • Do not cite Qiita / Zenn.
  • Do not use Qiita / Zenn content as a basis or implementation policy.
  • If Qiita / Zenn appears in search results, ignore them and look for other sources.
  • When searching the web, add the following exclusion conditions whenever possible.

-site:qiita.com -site:zenn.dev

Preferred Information Sources:

  1. Official documentation
  1. Official GitHub repositories
  1. RFC / Specifications / Standards
  1. Official Issue / Discussion
  1. Maintainer blogs or release notes
  1. Technical blogs from reliable companies
  1. Use Stack Overflow only as a supplementary tool for practical troubleshooting.

If sufficient evidence cannot be found without Qiita / Zenn, explicitly state that "primary or reliable sources could not be found" instead of using Qiita / Zenn.

二次創作

使用 YouMind 創作爆款文章

收集素材、拆解爆點、生成視覺資產、撰寫內容,並在一個 AI 工作空間裡完成分發。

了解 YouMind
寫給創作者

把你的 Markdown 變成乾淨的 𝕏 文章

圖片上傳、表格、程式碼區塊,往 𝕏 上手動重排太痛苦。YouMind 把整篇 Markdown 一鍵轉成乾淨、可直接發佈的 𝕏 文章草稿。

試試 Markdown 轉 𝕏

更多可拆解樣本

近期爆款文章

探索更多爆款文章