What problem does it solve? Manual literature reviews are slow, inconsistent, and often miss naming collisions or weak evidence. This Skill automates systematic literature research with subject disambiguation, citation network expansion, and evidence grading so every claim in the final report is traceable to a source. ## Core Features & Use Cases - Subject Disambiguation: Resolves identifiers (UniProt, Ensembl, ChEMBL, DrugBank) and detects naming collisions before searching, with domain-specific handling for genes, drugs, diseases, and general academic topics. - Systematic Search with Evidence Grading: Combines high-precision seed queries, citation network expansion, and collision-filtered broad queries across PubMed, ArXiv, Semantic Scholar, OpenAlex, and more; every claim is graded T1 (mechanistic) through T4 (mention). - Structured Deliverables: Produces a 15-section report with integrated biological model, testable hypotheses, and a JSON/CSV bibliography, plus a fast factoid mode for single-question verification. - Use Case: Ask "What does the literature say about ATP6V1A?" and receive a comprehensive report covering protein architecture, expression, disease links, research themes, and prioritized testable hypotheses with evidence grades. ## Quick Start Ask the agent to conduct a deep literature review on a gene, drug, disease, or research topic of your choice.
