What problem does it solve? Assessing whether scientific claims and research papers are trustworthy requires systematic evaluation of methodology, statistics, and bias—work that is easy to do superficially and hard to do rigorously. This Skill provides structured frameworks for critiquing research rigor so conclusions are proportional to actual evidence quality. ## Core Features & Use Cases - Methodology & Design Critique: Assess internal, external, construct, and statistical conclusion validity, including AI/ML-specific concerns like benchmark contamination and compute fairness. - Bias & Fallacy Detection: Identify cognitive, selection, measurement, and analysis biases plus 40 named logical fallacies with detection and mitigation strategies. - Evidence Quality Grading: Apply GRADE, Cochrane ROB, and evidence hierarchy standards, with a 20-point quick triage checklist for rapid paper screening. - Use Case: When reviewing a preprint claiming a new model beats baselines, use this Skill to check for cherry-picked benchmarks, missing ablations, p-hacking, and whether causal language is justified by the study design. ## Quick Start Ask the AI to critically evaluate the methodology and statistical validity of an attached research paper using the scientific-critical-thinking skill.
