Формулировка научных гипотез и экспериментальных предсказаний

★ 7.0 · research

hypothesis-generation is a Claude Code skill that turns observations and preliminary findings into transparent, testable research plans by strictly separating observations, hypotheses, mechanisms, predictions, and evidence. The workflow covers freezing the observation with provenance and uncertainty, framing the research question using PICO, PECO, or domain-specific constructs, generating candidate hypotheses alongside rival explanations, discriminating predictions, and preregistration-ready analysis plans. Bundled CLI tools run deterministically and locally, accepting bounded JSON, CSV, or Markdown with no network access, credentials, or external packages required. Mandatory safety gates prevent bypassing IRB, IACUC, biosafety, or dual-use review, and the skill never presents a hypothesis or mechanism as established evidence. Researchers who need to move from raw observations to rigorous, reproducible study designs will find it especially useful.