phy-paper-polish_skill

Overview

This skill helps agents review and polish research papers written in LaTeX, focusing on writing clarity, grammar, LaTeX best practices, and document structure. It provides targeted suggestions to improve readability, correctness, and reproducibility so papers are easier for reviewers to assess. The skill addresses both language issues and LaTeX-specific conventions to produce a cleaner, more professional manuscript.

How this skill works

The skill inspects LaTeX source (main.tex and subfiles) and the compiled output where available. It detects unclear sentences, passive voice, nominalizations, article misuse, punctuation issues, inconsistent section-title capitalization, and tense problems. For LaTeX it checks package choices, math typesetting (variables vs. text), numeric formatting, references (use of \autoref), bibliography setup, and recommended project layout (figures, tables, src, code). It returns concrete edits, examples, and patch-ready suggestions.

When to use it

• Before submitting a conference or journal paper to catch language and formatting weaknesses.
• When converting a draft into a camera-ready version to enforce consistent style and LaTeX best practices.
• During author collaboration to unify section titles, citations, and project directory layout.
• When preparing supplementary materials or code artifacts to ensure reproducible structure.
• To audit LaTeX sources for modern toolchain recommendations (LuaLaTeX, BibLaTeX) and portability.

Best practices

• Favor active voice, remove unnecessary words, and replace nominalizations with verbs for clearer prose.
• Pick one section-title capitalization style (title or sentence case) and apply it consistently.
• Use packages like siunitx for numbers, hyperref + \autoref for references, and BibLaTeX for bibliographies.
• Typeset variables in math mode with text subscripts (e.g., t_\text{max}) rather than italicized descriptive text.
• Organize project with a single entry file (main.tex) and separate directories for src, figures, tables, and code.
• Prefer modern engines (LuaLaTeX) for Unicode support and fewer font issues when possible.

Example use cases

• Polish wording and tighten the introduction to clearly state novelty, impact, and evaluation scope.
• Replace passive constructions and fix tense inconsistencies across methods and results sections.
• Convert manual numeric separators to siunitx \num{} and standardize measurement units.
• Switch bibliography to BibLaTeX and suggest fixes for incorrect citation commands.
• Reorganize a messy repository into main.tex, src/, figures/, tables/, and code/ for reproducibility.