auto-paper-improvement-loop

A two-round AI paper improvement loop autonomously reviews and fixes generated papers. It uses a secondary GPT model to evaluate writing, consistency, and structure, then applies corrections and recompiles the LaTeX document to enhance overall quality.

Yes, autonomous paper improvement integrates with latexmk for end-to-end LaTeX workflow support. The Skill processes your local LaTeX sources, applies AI fixes, and uses latexmk to recompile the final PDFs during each iteration of the two-round loop.

Yes, you can resume an interrupted AI paper improvement loop using state persistence. The Skill saves progress in PAPER_IMPROVEMENT_STATE.json, enabling you to recover and resume the two-round review and fix cycle exactly where it left off after interruptions.

The best way to track progress during an automated AI paper review is via PAPER_IMPROVEMENT_LOG.md. This log file documents each round, applied fixes, and outcomes, providing a clear record of the two-round improvement cycle and structural changes.

Yes, the automated paper improvement workflow requires a local paper directory with LaTeX sources. You must also have a local build tool installed, and you can optionally configure human checkpoint controls to approve AI fixes before application.

Yes, you can add human checkpoints to an autonomous AI paper editing workflow. The Skill includes optional human checkpoint controls, allowing you to review, approve, or provide guidance on AI fixes before they are applied to the LaTeX document.