This produces peer review quality analyses of academic papers with the kind of structured feedback you'd see at top venues like NeurIPS or Nature. It walks through the paper systematically, extracts claims and evidence, searches the literature to position the work, then generates a full review with rated methodology assessments, numbered strengths and weaknesses with specific section references, and questions for authors. The output follows standard review templates with metadata, executive summary, contribution list, and methodology scores across six dimensions like reproducibility and statistical rigor. Works with PDFs, arXiv links, or DOIs across any scientific domain. Honest take: the structure is more valuable than the critique itself, since it forces a thorough reading and prevents handwaving about vague concerns. Good for preparing your own reviews or understanding what referees will scrutinize in your submissions.
npx skills add https://github.com/bytedance/deer-flow --skill academic-paper-review