domains / science / hypothesis-generation / hypothesis-generation

hypothesis-generation

Skill card

combined

The copies overlap on the core scientific workflow (observation, literature search, competing hypotheses, quality evaluation, experimental design, and predictions), but they disagree on scope, safety constraints, and required outputs. The K-Dense-AI copy is the most rigorous, adding non-negotiable safety boundaries, a detailed object-distinction table, and a scope/safety gate that the others lack. The Vibe-Skills and brycewang-stanford copies are nearly identical but include a mandatory requirement for AI-generated schematics (via a separate skill) and specific tool usage (WebFetch/WebSearch) that the K-Dense copy does not. The aipoch copy adds specific LaTeX packaging and formatting requirements. Because the K-Dense copy's safety-first approach conflicts with the others' focus on visual deliverables and specific tooling, and because the others contain material steps (schematics, LaTeX) that the K-Dense copy omits, crowning one would hide real alternatives.

Repositories

Other copies

Combined skill

# Scientific Hypothesis Generation

## 1. Scope and Safety Gate
Before proceeding, confirm authorization for any sensitive or unpublished data. Ensure the work does not cross safety, ethics, or regulatory gates (IRB, biosafety, etc.). Do not present hypotheses as established facts or infer causation from association alone.

## 2. Understand the Phenomenon
- Identify the core observation or anomaly.
- Define scope, boundaries, and constraints.
- Clarify what is known vs. uncertain.

## 3. Literature Search and Synthesis
- Search PubMed (for biomedical) and general web sources.
- Synthesize current understanding, established mechanisms, and gaps.
- Identify contradictory findings or unresolved debates.

## 4. Generate Competing Hypotheses
- Develop 3-5 distinct, mechanistic hypotheses.
- Ensure they are distinguishable and draw on different levels of explanation (molecular, systemic, etc.).

## 5. Evaluate Hypothesis Quality
Assess each hypothesis for:
- Testability and Falsifiability
- Parsimony and Explanatory Power
- Scope and Consistency with established principles
- Novelty

## 6. Design Experimental Tests
- Propose specific experiments (lab, observational, clinical, computational).
- Define measurements, controls, and statistical approaches.
- Address potential confounds.

## 7. Formulate Testable Predictions
- Generate specific, quantitative predictions for each hypothesis.
- Identify conditions under which predictions hold.
- Specify observations that would falsify each hypothesis.

## 8. Report and Visualization
- Produce a structured report including the observation, hypotheses, predictions, and validation plan.
- **Mandatory Visuals:** Include at least 1-2 AI-generated schematics (e.g., hypothesis framework, experimental design flowchart) using the `scientific-schematics` skill or equivalent tool.
- If a LaTeX report is required, use the provided templates and ensure all references are compiled.