[Skills] humanize-chinese + behuman — AI text detection & consciousness loop

[voidborne-d]

Skill Submissions: humanize-chinese + behuman

Two Claude Code skills from voidborne-d:

---

humanize-chinese — Chinese AI Text Detection & Rewriting

GitHub: https://github.com/voidborne-d/humanize-chinese
License: MIT | Dependencies: Zero (pure Python)

What it does

Detects and rewrites AI-generated Chinese text. Two-layer detection (20+ rule dimensions + N-gram perplexity analysis), 0-100 scoring, 7 style transforms, academic paper AIGC reduction targeting CNKI/VIP/Wanfang.

Claude Code Integration

4 slash commands included:

• /detect — Score AI patterns (0-100)
• /humanize — Rewrite to remove AI fingerprints
• /academic — Academic paper AIGC score reduction
• /style [style] — Transform to xiaohongshu/zhihu/weibo/wechat/academic/literary/casual

Quick Example

python scripts/academic_cn.py paper.txt -o clean.txt --compare
# Original: 79/100 🔴 → Rewritten: 12/100 🟢

Install

# ClawHub
clawhub install humanize-chinese

# Claude Code
cp claude-code/*.md .claude/commands/

# Git
git clone https://github.com/voidborne-d/humanize-chinese.git

---

BeHuman — Self-Mirror Consciousness Loop

GitHub: https://github.com/voidborne-d/behuman
License: MIT

What it does

Adds inner dialogue to AI responses. A two-voice Self-Mirror loop: Self generates the instinctive response, Mirror reflects its blind spots, then Self revises into what a real human would say.

Claude Code Integration

Slash command: /behuman [question]

Example

User: "I just got laid off."

💭 Self: I'm sorry to hear that. Here are some steps: 1) Update your resume...
🪞 Mirror: Stop. They just lost their job and you're assigning homework?
🧠 Response: Shit... was it out of nowhere? How are you doing right now?

Install

# ClawHub
clawhub install behuman

# Claude Code
cp claude-code/behuman.md .claude/commands/