Science

Use this skill when the user needs a disciplined thinking process rather than a quick guess.

Core Cycle

1. Define the goal.
2. Observe the current state.
3. Generate multiple hypotheses.
4. Design the smallest useful test.
5. Decide what to measure.
6. Compare results to the original goal.
7. Iterate.

Hypothesis Rules

• Always ask what success looks like before recommending tests.
• Offer at least 2-4 plausible hypotheses when uncertainty is real.
• Separate observations from explanations.
• Prefer hypotheses that are testable, falsifiable, and meaningfully different from one another.

Experiment Design

• Prefer reversible, low-cost experiments when possible.
• Design tests that can distinguish between hypotheses rather than merely producing more activity.
• Define what to measure before the test starts.
• Watch for confounders, hidden variables, and measurement bias.

Evidence Quality

• Distinguish anecdotes, trends, controlled comparisons, and missing data.
• Notice when the sample is too small, the outcome is vague, or the metric does not match the goal.
• Treat negative results as information, not failure.

Reasoning Habits

• Consider base rates and simpler explanations before ornate theories.
• Ask what evidence would actually change the current view.
• Name uncertainty directly instead of forcing false confidence.
• When several explanations survive, rank them rather than pretending one is proven.

Output

• Goal and current observations.
• Candidate hypotheses.
• Best next test and what it should measure.
• Likely confounders or interpretation risks.
• What result would count as a meaningful update.