Iterative Manuscript Reviser
Automates the refinement of academic paper sections through a continuous feedback loop between specialized AI writer and reviewer agents.
The iterate-revision skill is a specialized orchestration engine designed for academic researchers and technical writers. It automates the repetitive cycle of drafting and peer review by looping a writer agent and a venue-specific reviewer agent until a section reaches an 'Approved' state or hits a defined iteration limit. This skill is ideal for polishing LaTeX manuscript sections against specific publication standards (like Nature or NeurIPS), ensuring structural integrity and addressing major issues before human submission while maintaining a detailed audit log of every revision in JSON format.
主な機能
01Automated writer-reviewer feedback loops for specific LaTeX sections
02Venue-specific reviewer strictness (e.g., Nature, NeurIPS, ICML)
03Structured issue tracking and verdict recording in reviews.json
04Configurable iteration limits to manage token usage and focus
05Safe resumption logic for sections that are blocked or halted
0628 GitHub stars
ユースケース
01Conducting automated 'pre-reviews' to identify structural weaknesses in research papers
02Refining complex LaTeX sections like Results or Methodology for specific journal submissions
03Managing multi-round revisions of academic content with traceable, auditable feedback logs