auto-paper-improvement-loop

Automate two-round LaTeX paper refinement with expert-style reviews and recompilation.

Updated Apr 1, 2026
One-click install
npx skills add https://github.com/Wenwen555/ARIS-LVLM --skill auto-paper-improvement-loop-wenwen555
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: auto-paper-improvement-loop
Source: https://github.com/Wenwen555/ARIS-LVLM/tree/main/skills/auto-paper-improvement-loop
Command: npx skills add https://github.com/Wenwen555/ARIS-LVLM --skill auto-paper-improvement-loop-wenwen555

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Automates two-round AI-driven refinement of a generated paper by performing expert-style reviews, proposing fixes, and recompiling to a polished LaTeX draft.

Core Features & Use Cases

  • Two-round AI-driven refinement using GPT-5.4 xhigh to identify logical, notation, and writing issues.
  • Stateful recovery and logging via PAPER_IMPROVEMENT_STATE.json and PAPER_IMPROVEMENT_LOG.md to ensure progress persistence.
  • Automated LaTeX recompilation with latexmk and preservation of intermediate PDFs (main_round0_original.pdf, main_round1.pdf, main_round2.pdf).
  • Optional human checkpoint for reviewing and approving fixes before automatic application.
  • Works with existing paper structure (paper/main.pdf, paper/sections/*.tex).

Quick Start

Provide the path to the paper directory (containing paper/main.pdf and paper/sections/*.tex) to start two rounds of automated improvement.

Frequently Asked Questions about auto-paper-improvement-loop

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I automatically review and improve a LaTeX paper?

Automatically reviewing and improving a LaTeX paper requires a two-round AI-driven refinement workflow that performs expert-style reviews, proposes fixes, and recompiles the sections into a polished draft.

What do I need to set up before running automated LaTeX paper refinement?

Automated LaTeX paper refinement requires an existing paper directory containing main.pdf and sections/*.tex files, a LaTeX toolchain like latexmk for recompilation, and AI model integration via Codex MCP.

How does state persistence work during AI paper revision loops?

State persistence during AI paper revision loops works by tracking progress and logging changes in a PAPER_IMPROVEMENT_STATE.json file, ensuring recovery and continuous refinement across multiple rounds.

Can I review AI-proposed paper fixes before they are applied to the LaTeX source?

Yes, you can review AI-proposed paper fixes before application through an optional human checkpoint, allowing you to approve modifications before the workflow automatically applies them and recompiles the LaTeX document.

Does automated LaTeX recompilation preserve intermediate PDF drafts during revision?

Automated LaTeX recompilation preserves intermediate PDF drafts during revision by saving versions like main_round0_original.pdf, main_round1.pdf, and main_round2.pdf alongside the final polished output.

What is the best way to fix logical and notation issues in generated research papers?

The best way to fix logical and notation issues in generated research papers is using a two-round automated AI review process that identifies writing flaws, proposes targeted fixes, and recompiles the LaTeX source.