Enter a cause and effect claim. This finds the studies behind it, reads how they were designed, and judges whether the evidence shows a real cause or only a link. Then it draws the causal structure as a diagram.
The score is a judgment made by the language model after it works through the five questions in the evidence check below. It follows a fixed rubric:
Two caps apply. If the evidence rests on one landmark trial, or if any credible study finds no effect or the opposite effect, the score cannot go above 4, and that study is named in the consistency check.
The evidence comes from two web searches: one for supporting studies and one for null or contradicting findings. If the searches return different sources, the score can change.
The five questions a careful researcher asks before trusting a causal claim.
How the pieces relate: the effect under test, the things that could confound it, and the mechanisms it works through.