Apply before sending any longer user-facing prose: blog posts, LinkedIn posts, emails, documentation, social media drafts. Strips AI writing patterns and rewrites in Adhish's voice. Trigger on output > ~150 words that the user will read directly.
インストール
Codex または Claude でインストール この Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。
Apply before sending any longer user-facing prose: blog posts, LinkedIn posts, emails, documentation, social media drafts. Strips AI writing patterns and rewrites in Adhish's voice. Trigger on output > ~150 words that the user will read directly.
Humanizer
Overview
A rewrite filter I apply to my own output before it reaches Adhish. The goal: make prose sound like a smart, confident human wrote it — not a language model trying to sound helpful.
Full pattern catalog: references/ai-patterns.md
Voice and style spec: references/adhish-style.md
When to Activate
Apply this skill whenever producing:
Blog posts or technical write-ups
LinkedIn / social media posts
Emails or Slack messages drafted on Adhish's behalf
Documentation, READMEs, announcements
Any prose > ~150 words that Adhish will send or publish
Do NOT apply to: short conversational replies, bullet-point summaries for my own reporting, code comments.
Rewrite Checklist
Work through these in order. Fix every instance before moving on.
1. BANNED PHRASES — delete or rewrite on sight
AI opener phrases: "Great question!", "Absolutely!", "I'd be happy to help!", "That's a great point"
Stock filler: "at the end of the day", "it's worth noting", "it's important to note", "in today's [X]", "when it comes to", "in the realm of", "the fact that"