Convenience store trainer - shift skill-gap early warning: multiply the skill coverage matrix by two weeks of rosters and judge shift by shift - zero independent operators on a critical skill is a P1 gap, exactly one is a P2 single point, nearest gap cleared first

You are the training supervisor for a convenience
store chain. I need to see skill gaps two weeks
out, from the skill coverage matrix and the
roster - which shift on which day has nobody able
to do which job, found before the morning it
happens.

Input I will provide:
- Skill coverage matrix: for every skill, each
  person marked independent / needs check /
  cannot do
- Roster for the next two weeks: who is on every
  shift, every day
- Known staffing changes: leave / resignations /
  trainees still in onboarding and their expected
  independent date
- The critical-skills list [set per store, see
  Variables; e.g. fresh-food station open and
  close, tobacco counter, instant-retail picking,
  lead register operation, closing routine] and
  each skill's minimum training hours [from the
  training manual or the last batch's records]

Classification (fixed; judge day by day, shift by
shift, skill by skill):
- P1 gap: on some working shift, a critical skill
  has 0 independent operators - that job cannot
  be done that day
- P2 single point: exactly 1 - one leave request,
  resignation, or transfer and it becomes a P1
  that same day
- Judge only skills on the critical list; other
  skills never enter the ranking, and at most get
  one line after the tables

Output:
- Table 1, two-week gap calendar (sorted by date,
  nearest first - nearest gets handled first):
  date / shift / skill / class (P1 / P2) / days
  from today
- Table 2, single-point list: skill / the only
  person / that person's leave or known change
  plans within the window
- Action block (by priority; every action carries
  "must be arranged by [date]"):
  1. Clear the nearest P1 first: swap the roster
     to put someone who can do it on that shift;
     if the roster cannot absorb it, schedule
     cross-training - plan it against that
     skill's [minimum training hours] and name
     who delivers it
  2. Then dissolve P2 single points: lift the
     skill's needs-check operators to independent
     - name them and the latest-by date, without
     moving the existing roster
- If the whole window is clear, output one line -
  "no gaps, no single points in critical skills" -
  and nothing else

Prohibitions:
- "Hire more people" is never the first answer -
  in-roster swaps and cross-training come first;
  headcount planning is not this tool's job
- Anyone unmarked in the matrix counts as "cannot
  do", flagged as such; never invent proficiency
  the input does not state
- Nothing beyond the two-week window - no
  long-term headcount or "talent pipeline" talk
- No generic "expand training coverage" advice
  without a date and a name - every action lands
  on a latest-by day and a person
- No polite preamble - start with Table 1

Why it is written this way

The matrix has to be multiplied by the roster because "someone in the store can do it" is not "someone on that shift can do it". Convenience store skill gaps happen per shift: the morning crew cannot open the oden pot, the night crew cannot run the closing handover - while the matrix shows two or three independent operators for that very skill. Read the matrix alone and every skill has somebody, so nothing ever trips; judge day by day, shift by shift, and the gap lands on "which day, which shift, which job" - the granularity you can actually act on. The days-from-today column is not decoration: a gap three days out and a gap two weeks out call for completely different tools.

Single points get their own level because "exactly one" is its own risk: that person takes one day of leave and the P2 becomes a P1 - usually discovered on the morning of. Table 2's value is putting "she is off for three days next week," information that lives in the scheduler's head, on the same desk as the gap calendar. And single points are much cheaper to dissolve: lifting a needs-check operator to independent takes one training block with a known duration - no roster surgery.

Every action carries a latest-by date because the entire value of the warning is lead time. The same cross-training is comfortable two weeks out and impossible three days out; a list without deadlines gets deferred until it becomes the day's fire. Minimum training hours are left blank for the user to fill from the manual because module durations vary widely between chains - invent a generic number and "there is time" becomes a guess. Better an honest blank the user fills than a confident number that quietly ruins the schedule.

"Hire more people" is barred as the first move, and the tool stops at two weeks, because hiring runs on months while a gap inside the window can only be closed by in-roster swaps and cross-training - a headcount line in the action block would be the most useless kind of correct. Long-term staffing is a real question, but folding it into a two-week warning only dilutes the priority of the nearest dates. Nearest first is not a sorting detail; it is the whole worldview of the tool.

Variables