Is the draft's subject your task, and what is missing?

The rule the check uses

The check first tests whether the task is a comparison. If the answer is none, the verdict is out of scope for the analysis part and the page names the framework part instead. If the answer is one system and the role is not a regulator or a test house, the verdict is reference only: the analysis material can inform a single-system assurance, but the draft's comparison subject is not the task. Any missing role, comparison answer, indicator, decision date or review date returns scope incomplete and names the missing answer, and a review date in the past returns scope incomplete because there is no live review to plan from. Otherwise the task is in scope. Four evidence answers are scored: a current practice statement, an accuracy result with its sample and class balance, an attack test at a stated attack potential, and an equality test across the groups the system will meet. The maximum is four. Four reads ready to test, two or three reads prepare the comparison, and under two reads prepare, with the missing evidence named in the actions. A decision date on or after the planning date reads plan for the published text, because a test run against a ballot draft should not be the final evidence for a decision taken after publication. The indicator is recorded for the reader and does not change the score, because the comparison plan is the same work at every level; it does change the pass mark the plan records. Nothing is assumed and nothing leaves the browser.

What it does not do

It does not score a system, test a face model, read a document or call anyone compliant. It does not store the answers, set a cookie or send an email. It places your task against the draft's public subject and tells you which evidence is missing, so that the next conversation with a supplier or a test house starts from the right question.