Can this skill design actual experiments?

Yes, it proposes specific experimental designs including what to measure, necessary controls, sample sizes, and potential confounding variables to address.

How does it handle literature reviews?

It utilizes WebFetch for targeted PubMed searches in biomedical research and WebSearch for general scientific papers to ground every generated hypothesis in current peer-reviewed evidence.

What scientific domains does this skill support?

This skill is domain-agnostic and can be applied across biology, physics, social sciences, and any other field requiring empirical inquiry and systematic testing.

Is this skill based on a specific research paper?

Yes, it is an implementation based on the Kosmos AI paper (arXiv:2511.02824) focused on AI for autonomous discovery.

What criteria are used to evaluate hypothesis quality?

Hypotheses are assessed for testability, falsifiability, parsimony (simplest explanation), explanatory power, scope, and consistency with established scientific principles.

Scientific Hypothesis Generation

Name: Scientific Hypothesis Generation
Author: jimmc414

byjimmc414

•

426

•

Data Science & ML

Generates testable, evidence-based scientific hypotheses and structured experimental designs from observations or literature.

This skill provides a systematic framework for autonomous scientific discovery, allowing Claude to transition from raw observations to rigorous, mechanistic explanations. It automates the process of literature synthesis using PubMed and web search, formulates multiple competing hypotheses, and evaluates them against strict scientific criteria like falsifiability and parsimony. By grounding every hypothesis in existing evidence and providing detailed experimental designs with specific predictions, this skill serves as an indispensable tool for researchers and data scientists aiming to accelerate the scientific method across any domain.

Key Features

01Automated literature synthesis via PubMed and WebSearch integration

02Quantitative prediction generation for empirical hypothesis testing

03Mechanistic hypothesis formulation from raw observations

04426 GitHub stars

05Structured experimental design including controls and statistical approaches

06Competitive evaluation of multiple explanatory models for any phenomenon

Use Cases

01Synthesizing existing literature to identify research gaps and testable explanations

02Planning mechanistic studies for biomedical or physical science projects

03Developing research directions from preliminary experimental data

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add jimmc414/kosmos hypothesis-generation

For use in Claude.ai and ChatGPT

Download Skill