What problem does it solve? Chinese academic drafts generated by AI often contain templated transitions, empty emphasis phrases, translation-flavored sentence order, unsupported strong claims, and formatting defects like raw Markdown bold or broken formulas. This reference provides concrete rules and checklists to detect and fix those issues while preserving research facts, data, and conclusion boundaries. ## Core Features & Use Cases - AI-trace removal: Identifies and rewrites high-risk patterns such as "首先/其次/最后" chains, hollow emphasis sentences, and promotional discussion paragraphs, replacing them with object-condition-evidence-boundary driven prose. - Formatting guardrails: Enforces rules for LaTeX formulas ($$...$$ blocks), structured tables for statistics and regression results, heading separation, and reference list layout. - Three-round self-check: Provides an evidence-based checklist covering structure, language, and factual/final-draft state, requiring counted evidence rather than simple pass marks. - Use Case: A graduate student pastes an AI-generated thesis introduction; the skill rewrites it into natural journal-style Chinese, downgrades unsupported claims like "显著优于" to bounded statements, and reports which AI traces were removed. ## Quick Start Paste your Chinese paper draft and ask the assistant to apply the writing-core rules to polish the text, remove AI-style phrasing, and run the three-round self-check with evidence.
