Rebuttal pipeline for conference paper reviews. Parses reviewer feedback, classifies concerns by severity/type, builds a per-reviewer response strategy, and drafts a venue-compliant rebuttal with placeholders for pending experiments. Supports follow-up rounds. Use when user says "rebuttal", "reply to reviewers", "respond to reviews", "rebuttal draft", or wants to answer reviewer comments for a conference submission.
Multi-dimensional research idea evaluation with iterative refinement and publication-quality proposal generation. Takes any research idea (free text, structured brief, paper extension, or file path), scores it across 8 dimensions (novelty, technical soundness, feasibility, significance, clarity, experimental validity, scalability, positioning), refines autonomously for 2-3 rounds, then generates a detailed proposal with method design (architecture diagrams, equations, pseudocode), experiment plan (benchmarks, baselines, ablations, supplementary experiments), and qualitative/quantitative analysis plans. Use when the user wants to "evaluate my idea", "research proposal", "refine this idea", "idea to proposal", "score my research idea", or "turn this idea into a paper plan".
Comprehensive topic-driven literature survey for AI/ML research. Searches arXiv, Semantic Scholar, and DBLP for recent papers (default: last 1 year), identifies research gaps, performs cross-domain exploration, and proposes 2-3 innovation directions with feasibility assessment. Use this skill when the user wants to survey a research topic, find related work, identify research gaps, or generate research ideas. Also use when the user says "literature review", "related work search", "find papers about", or "what's the latest on".
Paper-triggered literature survey and idea generation. Given a PDF, arXiv URL, or tweet containing a paper reference, analyzes the paper, searches for related work (including papers the input may have missed), performs cross-domain exploration, and proposes 2-3 innovation directions. Use when the user provides a specific paper and wants to understand its landscape, find extensions, or generate research ideas. Also triggers for "survey this paper", "what's related to this", "extend this work".
Automatic citation insertion for LaTeX manuscripts. Parses LaTeX source to find uncited statements (factual claims, method comparisons, background assertions, dataset references, numerical claims), searches for the correct papers via Semantic Scholar and DBLP, fetches official BibTeX (never fabricates), and inserts \cite{} commands. Returns modified .tex files and updated .bib file. Inspired by github.com/ZhangNy301/citation-assistant. Use when the user says "add citations", "find references for", "cite this", "my paper needs citations", or provides LaTeX files needing references.
End-to-end university assignment completion. Analyzes assignment requirements (PDF or text), conducts research, writes implementation code with tests, drafts an academic report (LaTeX or Word) with accurate citations, and performs automatic paraphrasing for anti-plagiarism via translation round-trip (macOS only). Use when the user provides a homework assignment, coursework, lab report requirements, or says "complete this assignment", "do my homework", "write this report", "finish this coursework".
Comprehensive academic paper draft feedback in ICML meta-review style. Analyzes correctness (equations, proofs, notation), motivation, methodology gaps, presentation quality, visualization improvements, and missing citations. Outputs structured feedback with overall assessment, strengths, critical/major issues, section-by-section comments, and a prioritized revision checklist. Use when the user wants "feedback on my paper", "polish my draft", "review my manuscript", "what should I improve", or "help me revise".
Conference-style academic paper peer review. Reads a paper PDF, assesses novelty, technical soundness, clarity, significance, reproducibility, and experimental design. Generates a structured review with summary, strengths, weaknesses, major/minor issues, questions, and scores. Supports NeurIPS, ICML, CVPR, ACL, AAAI, ICCV, ICLR formats with adjustable severity. Use when the user wants to "review a paper", "write a review", "assess this submission", "what are the weaknesses", or "generate a peer review".