學術
Research Skill
Academic research lifecycle for AI/ML papers — discover, deep-read, discuss/brainstorm, cite (BibTeX via DBLP→CrossRef→S2), write paper sections. Integrates Semantic Scholar / arXiv / AlphaXiv / OpenReview / DBLP, runs venue-quality gates (CCF, impact factor), and auto-injects expert knowledge from 85 bundled AI/ML domain skills across 21 categories (architectures, fine-tuning, RAG, inference serving, alignment, interpretability, evaluation, MoE, long-context, multimodal, etc.) — users do NOT invoke those directly; this skill routes to them. Trigger on "find papers on…", "what's new in…", "let's brainstorm research…", "is this paper any good?", "give me the bibtex", "help write related work", "reviews for <paper>", "citation graph of…". Not for product/framework how-tos, rewriting a doc into paper format, cleaning messy BibTeX, or vendor blogs / whitepapers / tutorials.