im-not-ai
AI가 쓴 한글을 사람 글처럼 윤문하는 Claude 스킬 — Korean AI-text humanizer: detects and rewrites translationese, mechanical parallelism, and 71 other AI tells
What is im-not-ai?
AI가 쓴 한글을 사람 글처럼 윤문하는 Claude 스킬 — Korean AI-text humanizer: detects and rewrites translationese, mechanical parallelism, and 71 other AI tells
Overview of im-not-ai
What it is
Humanize KR (im-not-ai) is a CLI skill that rewrites Korean text generated by AI (ChatGPT, Claude, Gemini, etc.) into natural human-written Korean without altering facts, claims, or figures. It detects and rewrites translationese, mechanical parallelism, and 71 other AI tells, categorized into 10 major categories × 85 sub-patterns (plus 1 hold), each classified by severity (S1/S2/S3).
Key features
- Detects and rewrites 10 categories of AI tells (translationese, excessive English citations, mechanical parallelism, AI-specific idioms, etc.) with 85 sub-patterns
- Three-path routing (light/standard/heavy) automatically selects the number of LLM calls (1/2/3+) based on text quality, reducing cost for well-written texts
- Supports multiple CLI tools: Claude Code, GitHub Copilot CLI, OpenAI Codex CLI, and Gemini CLI
- Four core principles: meaning preservation, evidence-based editing, genre retention, and prohibition of over-editing (warns at >30% change, stops at >50%)
- Configurable strict mode (
--strictor "precision mode" forces heavy path) - Open source under MIT license
Use cases
- Humanizing AI-generated Korean text for academic, business, or creative writing to remove AI-typical patterns
- Detecting and rewriting translationese and mechanical structures in Korean content
- Improving the naturalness of Korean text while preserving original meaning and data
Getting started
git clone https://github.com/epoko77-ai/im-not-ai.git
cd im-not-ai
./install.sh
After installation, use /humanize-korean in Claude Code or the corresponding command in other supported CLIs.
License and pricing
License: MIT. Open source (no pricing stated).
Based on the project's public README and repository metadata.