humanizer

Detects AI-generated writing patterns and rewrites text into natural human prose.

1|Updated Mar 18, 2026
One-click install
npx skills add https://github.com/psw7205/skills --skill humanizer-psw7205
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: humanizer
Source: https://github.com/psw7205/skills/tree/main/skills/humanizer
Command: npx skills add https://github.com/psw7205/skills --skill humanizer-psw7205

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve? AI-generated text carries recognizable tells—formulaic phrases, translation artifacts, sycophantic openers, and mechanical rhythm—that undermine credibility. This Skill diagnoses those patterns in English and Korean writing and rewrites the text so it reads as naturally human while preserving meaning, facts, and the author's voice. ## Core Features & Use Cases - Pattern Diagnosis: Scans text against a catalog of 30+ AI writing patterns (vague attribution, rule of three, chatbot framing, em dash overuse) plus a dedicated Korean rulebook covering translation artifacts, honorific mixing, and LLM-specific sentence skeletons. - Meaning-Preserving Rewrite: Rewrites flagged passages without inventing facts, changing claims, or altering names, numbers, dates, or quotations; flags unverifiable claims as unresolved rather than fabricating support. - Read-Only Review Mode: Returns diagnosis and suggestions without modifying source files when the user only asks for inspection. - Use Case: Paste a chatbot-drafted Korean blog post and ask for a rewrite—the Skill strips flattery openers, splits hedged agreement/rebuttal sentences, unifies speech levels, and returns a natural draft with a change summary. ## Quick Start Review this draft for AI writing patterns and rewrite it to sound natural while keeping my original meaning.

Frequently Asked Questions about humanizer

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I remove AI writing patterns from my text?▼

Submit the text and request a rewrite or humanize pass. The Skill scans for clustered AI tells like formulaic phrases, vague attribution, and chatbot framing, then rewrites flagged passages while preserving your facts, structure, and voice.

How to make Korean text sound less like AI translation?▼

Korean input triggers a dedicated rulebook covering translation artifacts like overused '~에 대해' and '~를 통해', English-style passives, formal noun padding, and honorific mixing. It processes these in a fixed priority order and keeps the original register intact.

Can it check my writing without changing the file?▼

Yes. Requests phrased as review, check, or analysis run in read-only mode, returning a diagnosis with suggested fixes and example edits. The source file is only rewritten when you explicitly ask for modification.

Does rewriting AI text change the facts or meaning?▼

No. The Skill preserves names, numbers, dates, quotations, claims, and argument direction, and never invents statistics, sources, or examples. Unverifiable claims are marked as unresolved rather than fabricated.

What are the limitations of AI writing detection?▼

Single markers like one em dash or clean grammar are not treated as proof; only clustered patterns trigger strong judgments. Formal documents keep their formal tone, and rewrites are rolled back if changes exceed roughly 30-50 percent of the text.