
from ai-collab-playbook210
Detects and removes common AI-generated writing patterns in English and Chinese, then injects natural human voice and rhythm for publication-ready prose.
This skill scans prose for predictable AI-writing patterns (filler phrases, formulaic structures, vague attributions, promotional language, overused constructs) and rewrites sections to sound natural and human. It supports both English and Chinese and provides practical rules, examples, and scoring guidance so an agent can both detect issues and produce humanized alternatives.
Use when text sounds robotic, overly promotional, or formulaic — for blog drafts, academic edits, marketing copy, or any content where a natural voice is required. Triggers include requests like "humanize this text", "remove AI writing patterns", or "make this sound more natural". Useful as a pre-publication edit step.
Well suited to writing and editing agents (Copilot/Codex-style assistants, Claude Code, Cursor) that can produce targeted rewrites and operate on bilingual content.
Pure instructional skill with no scripts — detects and removes AI writing patterns in English and Chinese prose. SKILL.md is well-structured with clear triggers, bilingual examples, and progressive disclosure to references/ and examples/ directories. Based on Wikipedia's Signs of AI writing guide. No security concerns whatsoever as it contains no executable code.
Clean instructional skill. Bilingual (EN/ZH) support is a nice differentiator. The scoring rubric inside the SKILL.md (directness, rhythm, trust, authenticity, density) is meta but useful. No scripts to test.