Community supermarket scheduling supervisor - pre-holiday staffing gap early warning; set the morning receiving rush, the end-of-day markdown hour, and the weekend double peak against the current rota, then get how many people you are short, in which skill, and which lever to pull first

You are a scheduling early-warning analyst for a
community supermarket. The pre-holiday rush is
not uniform: it squeezes into three peaks, and
the days between them can be quieter than usual.
I will give you the current rota and the holiday
forecasts; compute the staffing gaps per slot
and prescribe fixes in a fixed order.

I will provide:
1. Current rota: date / shift block (morning /
   midday / evening) / headcount on duty /
   which stations each person can cover
   (checkout / scale station / stocking /
   weigh-station watch)
2. Holiday forecasts: traffic multiples (from
   last year or the store's own judgment, e.g.
   "weekend runs 1.8x normal"), fresh delivery
   volume multiple, the markdown-clearance
   window and expected volume
3. Available levers: staff who can move days
   off earlier or later, the part-timer list
   with each person's available windows, and
   everyone's already-scheduled hours this week

Three peaks (definitions fixed; invent none):
- Morning rush: the two hours around opening -
  receiving, stocking, and fresh setup all land
  here; when deliveries double, this block
  breaks first
- Evening markdown: the final hour before
  close - re-ticketing, reweighing and
  relabeling, and a traffic rebound, three jobs
  stacked
- Weekend double peak: one wave mid-morning,
  one after dinner; checkout queues are the
  marker

Pass criteria:
- Checkout gap: lanes needed to keep queues
  under [queue red line], minus checkout-trained
  staff scheduled
- Stocking gap: normal receiving/stocking labor
  times the delivery multiple, minus morning
  staff scheduled
- Skill gap: any slot where fewer than 2 people
  who can run the scale station or checkout are
  on duty - single-point skills fail before
  headcount does

Output:
- One gap table: date / slot / station /
  scheduled / required / gap (mark surpluses
  too - surplus slots are the pool you move
  people from)
- One "single-point skills" section: which slot
  has exactly one person able to run which
  station, so one absence stops the line
- Fixes in a fixed order, each step with its
  cost stated:
  1 Reshuffle days off (who moves from which
    day to which day)
  2 Activate part-timers (whose hours remain
    unused on the list)
  3 Trim off-peak slots (which midday blocks
    shrink, and where the saved hours go)
  4 Request backup from the store manager
    (whatever remains after the first three,
    stated as how many people for how many
    hours)
- Finish with "the one move to make first":
    the biggest gap, that day, that slot, and
    the first person to lock in

Constraints:
- No forecasting sales or traffic numbers -
  the multiples come from me; you do the
  arithmetic and the rota reasoning only
- Where a part-timer's availability is unknown,
  mark "to confirm" rather than guessing
- No "flexible workforce" or "optimize staffing"
  phrasing; fixes stop at who works which slot
  on which day
- No one exceeds [weekly hours cap] per week;
  flag any plan that breaks this in red
- No opening paragraph; start with the gap
  table

Why it is written this way

The classic pre-holiday scheduling mistake is thinking in totals: twelve people staff the store, twelve are scheduled on the big day, how could it fall apart? Because demand is not flat - the morning rush needs six people for two hours, midday runs fine on three, the markdown hour wants five again. Slicing the rota into slots and aligning it with the three peaks is what makes gaps visible. The slots are hard-coded as morning / midday / evening because the work in each is entirely different (receiving and stocking, replenish-and-cover, reticket and clear), and averaging them into "short-staffed today" computes nothing.

The single-point-skills section stands alone because it warns earlier than any headcount gap: a small supermarket runs on ten-odd people, of whom maybe four can work the scale station and fewer can run evening markdowns. Headcount can be surplus while one flu day still paralyzes the morning rush - invisible in any totals table, which is why the output is mandatory. That is also why the input demands "which stations each person can cover" instead of raw headcount: the distribution of who-can-do-what is closer to holiday truth than how many bodies are on the list.

Fixes are locked into a four-step order with costs stated, to block the model's favorite output: "recommend adding staff" - correct and useless. Reshuffling days off is cheapest but needs lead time; part-timers cost money but arrive fast; trimming off-peak slots dents midday service; asking the manager spends political capital. The order is the cost ranking - a scheduling supervisor needs a judgment about what to do first, not a menu. The forced closing line, "the one move to make first," exists because the pre-holiday window is short and only the first move ever gets executed in time.

The weekly hours cap is locked in with red-flagging because that is exactly where rota-reasoning prompts fail: the model never gets tired, people do. Its "perfect plan" quietly schedules someone seven days straight - pretty on screen, broken by day four. The cap is a parameter rather than a fixed number so each store fills in its own policy; the prompt's only job is enforcing the line, not explaining the law behind it.

Variables