humanize-academic-writing

This skill transforms AI-generated academic drafts into more natural, human-sounding scholarly prose for the social sciences. It targets a real pain point for non-native English speakers who have genuine research and ideas but produce text that reads as 'too perfect,' mechanical, or repetitive after drafting with AI tools. It opens with an explicit academic-integrity statement framing acceptable use as improving the expression of one's own original ideas, not generating or disguising unowned work.

The workflow runs an AI-pattern analyzer script (ai_detector.py) that flags repetitive sentence structures and lengths, overused transition phrases (Moreover, Furthermore, Additionally), abstract/vague scaffolding, mechanical paragraph transitions, low vocabulary diversity (type-token ratio), excessive passive voice, and consecutive-sentence similarity, producing a probability score and per-paragraph issues. It then applies targeted rewriting strategies: varying sentence rhythm (burstiness), reducing abstract scaffolding, eliminating mechanical connectors, adding scholarly voice with appropriate hedging and critical engagement, and grounding claims in specific theories, scholars, and concrete examples. Each rewrite is delivered with original text, revised text, and a short rationale, and the skill codifies principles like perplexity, specificity over abstraction, and natural flow, plus discipline-specific language notes for fields such as sociology.

Target users are social-science researchers and students, especially non-native English writers in sociology, anthropology, political science, education, and psychology, who want to improve writing quality and reduce obvious AI markers in drafts of their own work. Typical use cases include revising an AI-drafted paper section, tightening vague passages, and learning better academic writing patterns.