File 11, executable

Score something

Weights and point ceilings come from the corpus, so the number here is the number the documentation describes. A dimension you leave unscored is reported as missing — it is never counted as a zero.

Finding quality score Opportunity priority score Authority score Lead qualification score

Finding quality score

FindingQuality = weighted mean of dimensions x 20 · output 0-100

Evidence (30%) 0 = no support · 3 = credible single source/corroborated secondary · 5 = E4–E5, reproducible and current

Entity resolution (15%) 0 = unresolved · 3 = probable match · 5 = exact legal/domain/person-role match

Recency (15%) 0 = obsolete/unknown · 3 = relevant but aging · 5 = current for the decision/window

Directness (15%) 0 = rumor · 3 = proxy · 5 = direct observation/measurement

Independence (10%) 0 = copied origin · 3 = uncertain origins · 5 = independent origins

Contradiction check (15%) 0 = none · 3 = basic check · 5 = strong alternatives actively tested

Reset

What this does not do

A high score never overrides a hard gate, and nothing here stores what you type. The calculation happens on this request and is gone when the response ends.

How the rubrics are defined GET /v1/scoring/finding