Schedule Builder

A schedule is its columns. Adding one after a hundred rows are filled means going back through every item, which is why schedules end up with the columns somebody happened to start with and a set of notes doing the work the missing columns should have done.

Workflow

Step 1. Establish the type and where the items come from

Ask, in one message:

1. What kind of schedule, and is there an office standard column set? Paste it if so.
2. Where do the items come from: a model, plans, a list, or all three?
3. Which specification sections will reference it, and do they ask for anything by name?
4. What phase is the set in? A DD schedule and a CD schedule carry different columns and pretending otherwise fills a lot of cells with placeholders.
5. Does anything have to match an external format?

Step 2. Settle the columns before any row

Present the proposed column set and hold it against three tests:

Column	Why it is here	Who fills it	Needed at this phase?

• What does the specification look for? A schedule that does not carry what the spec references is a schedule the contractor cannot use.
• What does the project actually have? A column that will be identical in every row is a general note, not a column.
• What will be needed later? Adding a column at CD means revisiting every row. Adding an empty one now costs nothing.

Get explicit agreement on the columns before proceeding. This is the step the skill exists for.

Step 3. Fill only what you have

One row per item. Every cell is either a value you supplied or the word not supplied. There is no third state and nothing is defaulted.

Where a value repeats across many rows, say so and ask whether it belongs in the schedule at all or in a note. Where a value varies unexpectedly within a type, flag it; that is usually an error in the model rather than a real variation.

Report the count of not supplied cells by column. A column that is mostly empty is either premature for this phase or unnecessary.

Step 4. Reconcile against the drawings, both directions

Two lists, and both matter:

• Scheduled but not drawn. Items in the schedule with no location on a plan.
• Drawn but not scheduled. Items on the plans with no schedule row. This is the more common and the more expensive, because it gets found on site.

Then check the marks themselves: duplicates, gaps in a numbering sequence, and any mark that does not follow the set's convention. A duplicated door mark is found by the hardware supplier, late.

Step 5. Report

The schedule, then:

• not supplied counts by column.
• The two reconciliation lists.
• Mark problems: duplicates, gaps, non-conforming.
• Any column the office standard lacks that this project needed, named, so the standard can be fixed rather than worked around again.

Step 6. Save

Ask where and in what format. CSV where it goes back into a model or a spreadsheet, markdown where it is read.

Rules

• Never default a value. not supplied is the only alternative to a value you were given.
• Never carry a standard hardware set, rating, or finish.
• Settle the columns and get agreement before filling any row.
• Reconcile in both directions, every run.
• Report not supplied counts by column, not as one total.
• Flag a column that would be identical in every row.
• Never check compliance here; name the skill that does.

Anti-patterns

• Starting to fill rows before the columns are agreed.
• Defaulting a rating, a hardware set, or a finish because it is typical.
• Reconciling only from the schedule to the drawings.
• Carrying placeholder values that look like real ones.
• Adding a column at CD and filling the earlier rows from assumption.
• Building a schedule the specification does not reference.
• Treating a duplicated mark as a formatting issue.

Resources

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