Skillon
AI Humanizer preview
Speak this trigger phrase to your agent to launch the workflow
/de-ai-text
Skillon Pick

AI Humanizerv1.0.0

Use this Skill: Rewrite AI-sounding English so it reads genuinely human

00Instant ZIP · one-command CLI installFree for membersVerified

$3.99

Subscribe — use every SkillOne subscription unlocks the full catalog, including this Skill.

An AI-humanizing workflow built on how AI detectors actually work (perplexity / burstiness / lexical diversity / structural fingerprints). Locks what must not change, diagnoses AI signals with a 10-point scorecard, removes them with 14 rewriting techniques, then injects traces of real human authorship plus a triple self-check. For blog posts, essays, marketing copy, and scripts. Call it when the user says "too AI-sounding", "make it sound human", "humanize this", or "lower the AI score".

Updated 2026-09-30

Runs on

Claude CodeCodexCursorGitHub Copilot

Similar skills

What you'll get

  • Rewritten full text: a publish-ready revised version
  • Change log: 5–8 entries in 'what changed → which technique → why' form
  • Remaining risks: spots where signals still run strong + a list of details for you to fill in
  • Includes 6 reference manuals: AI-cliché library, detector-signal deep dive, synonym-swap table, self-test flow
de-AIrewritinglower AI scoresocial mediapapers

A good fit if you

  • De-AI-ing posts before publishing on WeChat Official Accounts / Xiaohongshu / Zhihu
  • Papers, assignments, reports that need more natural expression
  • Voiceover scripts and video narration that sound like a robot reading
de-AIrewritinglower AI scoresocial mediapapers

Might not fit if you

  • Needing a guaranteed score on a specific detector (free detectors are unreliable for Chinese — nobody can promise a score)
  • English text rewriting (this skill is designed for Chinese)
  • Fabricated data or made-up cases (conservative mode forbids fabricating facts)
de-AIrewritinglower AI scoresocial mediapapers

About this skill

You say:Say /de-ai-text, and rewrite this passage so it sounds human:
/de-ai-text

An AI-humanizing workflow built on how AI detectors actually work (perplexity / burstiness / lexical diversity / structu…

AI Humanizer example

FAQ

No — and you shouldn't trust any tool that promises a specific score. Free detectors are generally unreliable for Chinese (even the US Constitution was once misflagged as AI). This skill's goal is to write something that genuinely reads human: get the texture right and the score follows, and every edit comes with a change log so you know why each change was made.