What problem does it solve?
Manually polishing academic papers to meet peer review standards is slow, inconsistent, and often misses critical structural, theoretical, or visual flaws that reviewers flag during submission.
Core Features & Use Cases
- Bias-Guarded Iterative Review: Runs 2 rounds of GPT-5.5 xhigh paper review with fresh reviewer threads each round to avoid confirmation bias, scoring papers on theoretical rigor, claim-evidence alignment, and visual quality.
- Automated Fix Implementation: Parses reviewer weaknesses by severity (critical > major > minor) and applies targeted fixes to LaTeX source, with optional edit whitelists for resubmit and camera-ready constraints.
- Built-in Quality Safeguards: Includes restatement regression tests to prevent theorem drift across rounds, format compliance checks for page limits and duplicate labels, and optional adversarial kill-argument checks for theory-heavy papers.
- Use Case: A researcher with a compiled MICCAI 2025 paper can run this skill to automatically catch overclaims, fix notation inconsistencies, improve figure layout, and boost their paper's review score before submission.
Quick Start
Run the auto-paper-improvement-loop skill on your compiled LaTeX paper directory to start the 2-round review, fix, and recompile process.
