What problem does it solve? Turning a raw observation or preliminary data into rigorous, testable scientific hypotheses is hard: explanations tend to be vague, unfalsifiable, or cherry-picked. This Skill structures the process of proposing distinct mechanistic explanations, grounding them in the literature, and designing experiments that tell them apart. ## Core Features & Use Cases - Competing Hypothesis Generation: Produces 3-5 genuinely distinct mechanistic hypotheses, each evaluated against testability, falsifiability, parsimony, explanatory power, scope, consistency, and novelty. - Literature Grounding: Applies PubMed and web search strategies (reviews first, then primary research, then citation mining) with source-quality criteria and citation organization. - Experimental Design: Matches designs to claims (RCT, cohort, case-control, in vitro, computational) with controls, blinding, power analysis, and confound mitigation, plus falsifiable and distinguishing predictions. - Structured Reporting: Ships a LaTeX report template with color-coded hypothesis, prediction, and comparison boxes, plus a formatting guide for main text and appendices. - Use Case: Given an unexpected experimental result, generate three competing mechanistic explanations, a discriminating experiment for each pair, and a formatted report with 50+ cited references. ## Quick Start Use the data-analysis-hypothesis-generation skill to propose competing testable hypotheses for this observation and design experiments to distinguish them.
