Skip to content

Field types

Each field in a collection has a type. The type controls how values are stored, how the chat assistant fills them in, and how the field renders in the app.

Free-form text. Use it for anything that doesn’t fit another type — descriptions, notes, names.

A numeric value. Use it for amounts, counts, ratings.

A date in YYYY-MM-DD form. Renders as a date picker on the item detail screen.

A place — typically lat/lon plus a label. The assistant resolves addresses and place names to coordinates and stores both.

A picture attached to an item. The field’s value is the id of an image document in your library. When you attach a photo to a field — from the camera, your photo library, or one the chat assistant generated — Moonjar captures the image into Documents first, then writes that document’s id into the collection field. The detail screen reads the id back and renders the image inline.

True or false. Renders as a toggle. Accepts a few common synonyms on write (yes/no, on/off, checked) and normalises them to true or false.

A single choice from a fixed list of options. You define the options when creating the field — “low, medium, high”, for example. The assistant only writes values that match one of the options (case-insensitive); anything else is rejected with a clear error.

A pointer to a single item in another collection you own. Used to link related items — a book reference on a reading list collection points at a row in your books collection. See References between collections.

Like reference, but points at any number of items in another collection. Use when an item can relate to many others — a trip that visits multiple places, for example.

When you add an item by chat, the assistant fills each field according to its type:

  • Text and number are passed straight through.
  • Date is parsed and normalised to YYYY-MM-DD. Vague dates (“last March”) become exact ones using your timezone.
  • Boolean accepts yes / no / true / false and synonyms.
  • Selection has to match one of the declared options. The assistant uses the option list as part of its prompt so it knows what’s valid.
  • Reference and multi-reference require linking to specific items. The assistant looks them up by name first, then asks if it can’t find a clean match.
  • Location can be set from a place name — the assistant geocodes it.
  • Image can be a URL, an existing photo in your library, or a picture the assistant generated.
  • A field’s type is set when you create the field. To change it, you delete the field and add a new one (the assistant can do this for you).
  • Selection fields with no options will fail validation — you must declare at least one option.
  • Reference and multi-reference fields target a collection you can access. The app’s collection picker shows the collections you own and the ones shared with you through a group; pick from there.