Design Occupancy Load

Turn a list of spaces into an occupant load, one space at a time, with the factor you supplied next to every number. Divide area by factor is the same arithmetic everywhere. The factors are not, so the factors come from you.

Workflow

Step 1. Establish the code and the basis

Ask, in one message:

1. Which code governs this project, and which edition?
2. What is being counted: a whole building, one floor, or specific spaces?
3. What area basis are your numbers on, gross or net, and do you know which basis each factor requires? The two differ enough to halve or double a count, so I will ask again per factor if you are unsure.
4. How will you supply the factors? Type them, paste the table row, or name a file I should read.

Record the code and edition. They go on every output row, so the table can be read next year without guessing the cycle.

Read the profile first, if there is one. If docs/jurisdiction.md exists, read its occupancy load section before asking anything else and say which use and function combinations it already answers. Match on the Key column, which is where the area basis is recorded: a factor keyed to net area is not the factor for a gross measurement, and swapping them silently halves or doubles the count. Those combinations are not asked again. A row marked not cited stays not cited in the output. Where the file is absent nothing changes, and jurisdiction-profile is what writes it.

Step 2. Collect the space list

For each space or group of like spaces, ask for:

• Name or room number
• Use, as your code classifies it
• Function within that use where the code splits it (seating layout, sales versus storage, and so on)
• Floor area, and whether that figure is gross or net
• Fixed seating present, yes or no, and the seat count if yes

Never fill in an area or a use. Anything unstated is missing and is reported that way.

Step 3. Collect one factor per unique use and function

Work through the unique combinations, not the spaces, so a factor is asked for once and reused. For each, accept exactly three kinds of answer.

1. A typed factor. Record it, and ask for the clause it came from and the area basis it applies to.
2. A pasted table row. Read the factor and the basis out of it, and quote the row back so it can be checked.
3. A file to read. Read it, quote the row, cite file and line.

If none arrives, mark every space in that combination unverified and continue. Do not offer a factor, do not search for one, and do not treat a familiar number as exempt.

When a factor arrives that docs/jurisdiction.md did not hold, offer to write it there with its clause and its area basis, so the next run does not ask. Never write a row the user did not supply.

Ask separately for the rounding rule, since codes differ on whether a fraction rounds up, and on whether rounding happens per space or on the total.

Step 4. Compute and report

Space	Use	Area	Basis	Factor	Basis matches	Load	Cited as	Status

• Basis matches: whether the area basis you gave for the space is the one the factor requires. A mismatch is reported, never silently reconciled.
• Load: area divided by factor, rounded by the rule you supplied. Blank when the row is unverified.
• Cited as: code, edition, clause, or file and line, or not cited.
• Status: counted, unverified (no factor), missing (no area or use), seats (fixed seating, count supplied), or basis mismatch.

Then subtotals by use group, a total of the counted rows, and a separate count of what was not counted.

Close with:

Factors in this table came from you, not from me. I did the arithmetic and supplied no code values. Unverified rows are not in the total.

And a reminder that an exit, stair, or corridor serving several storeys needs the cumulative load, which this run did not apply.

Step 5. Save or hand off, if asked

Ask whether to write the table to a file and where, and whether to emit the subtotals as JSON for a downstream check. Write it directly. There is no template and no script.

{
  "code": "<code and edition as given>",
  "computed_at": "<ISO timestamp>",
  "per_space": [],
  "per_use": {},
  "counted_total": 0,
  "unverified": 0
}

Rules

• Never supply a load factor. Not from memory, not from a search, not from a figure in this file. There are none in this file.
• Never reconcile a basis mismatch quietly. Gross against net is the error that halves or doubles a count while looking correct. Report it and ask.
• A space is counted only with both an area and a factor. Everything else is unverified or missing, and both are reported next to the total.
• Round by the rule supplied, not by convention. Ask whether it applies per space or to the total.
• Fixed seating is a seat count from you, never an area divided by a factor.
• Say on every run that cumulative loads for multi-storey egress components were not applied.

Anti-patterns

• Filling a factor from recall because the use is a common one. Common factors are the ones that get carried across an edition change unnoticed.
• Searching the web for a table row and using it as the factor. Search finds which code governs; it does not quote it reliably, and a row that is nearly right reads exactly like one that is right.
• Adding unverified rows into the total to make the number look finished.
• Reusing one factor across two functions of the same use because they seem close. Ask which row applies.
• Converting between square feet and square metres, or rounding an area to a neater figure before dividing.
• Bundling a factor table into this skill later. It has none on purpose, and a bundled table goes stale silently while still looking authoritative.

Resources

None. This skill is one file. Output is written directly at the path you name.