moai-domain-humanize

Detects and removes AI-writing tells from Korean, English, Japanese, and Chinese text while preserving meaning.

1.2k|222|Updated Sep 16, 2025
One-click install
npx skills add https://github.com/modu-ai/moai-adk --skill moai-domain-humanize
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: moai-domain-humanize
Source: https://github.com/modu-ai/moai-adk/tree/main/internal/template/templates/.claude/skills/moai-domain-humanize
Command: npx skills add https://github.com/modu-ai/moai-adk --skill moai-domain-humanize

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve?

AI-generated text carries recognizable patterns—formulaic connectors, inflated significance, cliché openers, and template structures—that make it read as machine-written. This Skill post-edits (윤문) such text to read as human-authored without changing facts, figures, or the author's stance.

Core Features & Use Cases

  • Multilingual tell catalogues: Language-specific detection categories for Korean, English, Japanese, and Chinese, each with prose and marketing-copy layers and before/after rewrite examples.
  • Shared severity and grading model: S1/S2/S3 severity tiers gate removal by occurrence count, and A/B/C/D quality grades decide whether output ships, gets a second pass, or needs human review.
  • Over-editing guardrails: Prose mode warns at >30% change and halts at >50%; copy mode enforces fact-anchor preservation so numbers, dates, prices, and proper nouns survive verbatim.
  • Use Case: Paste an AI-drafted English blog post and receive a humanized version plus a change report listing detected tell categories, the quality grade, and the estimated percent changed.

Quick Start

Humanize this AI-generated draft and report which tell categories you removed and the final quality grade.

Frequently Asked Questions about moai-domain-humanize

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

FAQPage Schema
How do I make AI-generated text sound human?

Run a humanization pass that detects catalogued AI tells—formulaic connectors, cliché openers, inflated significance—and rewrites only the flagged spans. This Skill supports Korean, English, Japanese, and Chinese with per-language tell catalogues and before/after examples.

What AI writing tells does this detect in English text?

It detects ten English categories including focal vocabulary like delve and tapestry, negative parallelism, rule-of-three tricolons, signposting fillers, transition overuse, cliché openers, vague attribution, and formatting tells, each with severity-based removal thresholds.

Does humanizing text change its meaning or facts?

No. Meaning preservation is the top rule: facts, numbers, dates, names, and the author's stance are anchored in an invariant ledger before editing, and a post-edit delta audit rolls back any edit that drifts from them.

Can it humanize marketing copy and landing pages?

Yes. A copy mode covers headlines, CTAs, landing pages, and slide titles with language-native copy-layer catalogues. The change-rate guard is replaced by a fact-anchor guard, so headlines can be freely rewritten while numbers and promises stay intact.

What are the limitations of AI text detection here?

Automated AI detectors are unreliable, especially on CJK polite registers, so this Skill is a pattern-based editing tool rather than a detection oracle. Change-rate percentages are LLM estimates, not computed metrics, and borderline cases are treated conservatively.