The week, costed as you build it

The same eight hours cost more if they are the fiftieth hour of the week than if they are the eighth. So the engine walks each person's week in order and books the premium to the day it was earned — not to the end of the week, where it would tell you nothing about which shift to move.

Daily and weekly thresholds do not stack. An hour already paid as daily overtime is not counted a second time toward the weekly one; the result is the larger of the two, not the sum. Getting that wrong overstates a week's cost by a margin that is easy to miss and expensive to explain.

Hours worked plus hours still scheduled says who crosses on Friday. The row total — 47h · $885 — turns amber and says on hover that three of those hours are at a premium, costing $27 more than straight time.

Four hours, because that is shorter than most shifts: close means the next one will probably cross, rather than might. A margin the length of a full shift would light up half a roster every week and stop being read.

A default budget that holds for every week, and a figure that overrides a single one — because December is not a normal month and pretending otherwise means ignoring the bar for six weeks. Day columns carry the day's cost, person rows carry the week's, and each day is drawn against the same day last week. Under five per cent is not drawn at all: a day that moved by the price of one short shift has not moved.

The person's own hourly rate first, then the role's default. An open shift is priced at the role's rate, because when it fills it will spend that money. A shift with no rate anywhere is left out of the total rather than counted as zero, and the bar tells you how many were left out.

Seeing wages is a permission, not a display toggle. A scheduler builds the week, answers requests, edits the roster and sees every warning — and gets no money at all. Rates, the budget and last week's figures are null before the page is rendered, so there is nothing for a browser's developer tools to find.