First you need to have a model. For example, here is a document with a couple of objects. I did Arch stuff here, but it doesn't need to be Arch, it can be anything.
Then you press the Arch Schedule button. You get a task panel like this. It is pretty wide, so you'll need to widen the task panel to be comfortable.
Then you can fill line by line. Each line is a "query" and will render one line in the spreadsheet. Press the Add button to add a new line, and double-click each cell from that lien to fill in the values. Values are:
- Description: A description for this query. The Description column will be the first column of the resulting spreadsheet. The description is mandatory to have a query performed. If you leave the description cell empty, the whole line will be skipped and left blank in the spreadsheet. This allows you to add "separator" lines.
- Value: This is the real query that you want to perform on all the objects selected by this query. It can be two kinds of things: either the word count (or Count or COUNT, it's case-insensitive), which will simply count objects, or retrieve ans sum a property, for example object.Shape.Volume or object.Length or even object.Label. The name you use before the first dot (object) can be anything, you could also write x.Shape.Volume. The rule is: what comes after the first dot will be retrieved from each object selected by this query, if possible (object that don't have the required property will be skipped), and the result will be added together. For example, if you use object.Shape.Volume, you will get the sum of all volumes of all objects selected by this query.
- Unit: An optional unit to express the results in. It's up to you to give a unit that matches the query you are doing, for example, if you are retrieving volumes, you should use a volume unit, such as m^3. If you use a wrong unit, for ex. cm, you'll get wrong results.
- Objects: You can leave this empty, then all the objects of the document will be considered by this query, or give a semicolon (;)-separated list of object names (not labels). If any of the objects in this list is a group, its children will be selected as well. So the easiest way to use this feature is to group your objects meaningfully in the document, and just give a group name here. You can also use the Selection button to add objects currently selected in the document.
- Filter: Here you can add a semicolon(;)-separated list of filters. Each filter is written in the form: filter:value, where filter can be (it's case-insensitive too): Name, Label, Type, Role, or !Name, !Label, !Type, !Role. Fro example: name:door;type:window will filter the objects we got from the step above, and retain only those whose name contains "door" AND the type (returned by Draft.getType) is "wall". Everything is case-insensitive. Filters that begins with ! are inverted. For example, !name:wall will retain only objects that DON'T have "wall" in their name. "Role" is a property that all Arch objects have.
So we can build a list of queries like this:
After that, press OK and a new Schedule object is added to the document, which contains a result spreadsheet:
By double-clicking the Schedule object, you get back to the task panel and change the values. By double-clicking the spreadsheet itself, you get the results in 3 columns: description, value, unit (if applicable):
The spreadsheet can then be exported to csv normally, from the Spreadsheet workbench.
That's more or less it. No more fiddling with prices, no more complex stuff, all that you would now do in a better spreadsheet app, outside of FreeCAD. The idea here is that you can "drop-in and drop-out" pieces of spreadsheets between FreeCAD and another spreadsheet. For example, you build a price list in LibreOffice. You reserve 5 columns at the end of your prices list that will be exported to csv and contain just the things we need here above. Then, you import back the result from FreeCAD into your "quantities" column, and bingo, your price list is done

Of course there is much room for improvement. The filters list is somewhat limited to Arch stuff, and could be widened a lot, the objects selection system is not 100% good yet, one could think of auto-syncing with libreoffice, etc. But I think now this tool is on the right track, it can basically get "anything" from the document.