Analyze computer vision or graphics papers from local LaTeX source files/directories or Markdown documents, optionally incorporating public OpenReview reviews, and write a Chinese evidence-grounded overview.md.
Extract academic source materials into a local ./papers workspace. Use when the user provides a local PDF that is clearly a paper, technical report, proposal, preprint, or other academic document, when the user provides an explicit arXiv ID or arXiv URL, or when the user provides an OpenReview forum ID or OpenReview forum URL. Download and unpack arXiv LaTeX sources for explicit arXiv inputs, download public OpenReview submission PDFs plus metadata for explicit OpenReview inputs, and handle MinerU for non-arXiv PDF inputs in manual, background, or inline mode.
Review academic papers or manuscripts from local LaTeX source files, LaTeX project directories, or Markdown documents, and write a Chinese evidence-grounded peer-review `review.md` beside the input. Optionally incorporate OpenReview forum, review, rebuttal, or discussion materials after first resolving an OpenReview-capable Python interpreter that can `import openreview`. Use when the user asks to 审稿, review, or act as a reviewer for a paper/manuscript. Do not use for code review, thesis review, or NSFC proposal review.