Paste an abstract — or just a DOI, PMID, or PubMed link — and get a 14-point scorecard built on the Good Study Framework. The core question, always: does this show X causes Y, or just that they appear together?
Scores are computed deterministically from the text and registry metadata (PubMed publication types, retraction flags). A score is a structured read, not a peer review — it can't see what the paper doesn't say. Inspired by Emily Oster's data methodology.