What problem does it solve? Academic drafts often fail not because of grammar but because of broken argument structure, overclaims, or prose that does not match journal expectations. This Skill diagnoses and fixes those problems by polishing, restructuring, or translating manuscript prose into concise Nature-leaning English without inventing scientific content. ## Core Features & Use Cases - Axis-based fragment loading: Detects paper type (research, methods, hypothesis, algorithmic, review), section (abstract, intro, results, discussion, conclusion, title, methods), language (English or Chinese-to-English), and target journal, then loads only the relevant guidance fragments from static/ via manifest.yaml. - Structural and phrase-level polishing: Applies claim-evidence-boundary logic, hourglass structure, hedging calibration, overclaim control, and Academic Phrasebank-derived phrase families from references/. - LaTeX layout repair: Fixes typesetting problems such as loose float pages, stranded section headings, "Float too large" errors, and sparse Supplementary Information using references/latex-layout.md. - Use Case: Paste a Chinese-drafted discussion section and ask for publication-quality English; the Skill detects zh-to-en, loads the discussion and language fragments, rebuilds the logic, and returns polished prose with revision notes. ## Quick Start Polish this abstract into Nature-style English and explain the structural changes you made.
