What problem does it solve?
This Skill solves the common pain point of AI-generated academic and serious nonfiction prose reading templated, robotic, or obviously AI-written, while eliminating the risks of over-editing human-quality text, breaking scholarly register, or inventing facts during manual rewriting.
Core Features & Use Cases
- Mode-aware rewriting: Supports two distinct modes: academic for scholarly papers, theses, and research reports, and popsci for serious science journalism and explainers, with mode-specific rules to preserve appropriate register and legitimate craft (e.g., rhetorical questions in popsci are preserved, while treated as register slips in academic mode).
- False-positive safe abstain protocol: Automatically leaves human-quality prose unchanged if it already reads naturally for its target mode, only proceeding with rewriting when specific, removable AI writing signals are identified.
- Independent quality validation: Uses a blind judge rubric (separate from pattern detection) to score rewrites, ensuring register is preserved, no facts are invented, and the text reads as genuine human writing.
- Use case: Ideal for cleaning up AI-drafted thesis chapters, literature reviews, research reports, or science explainers that feel templated or obviously AI-generated, without losing formal academic tone or factual accuracy.
Quick Start
Provide the AI-generated academic or serious popular science text you want to humanize, specify whether it is academic or popsci mode, and ask the Skill to rewrite it while preserving all original facts and the target mode's register.
