How It Fits Together
You add one dependency. It gives you three modules — Auth, Inventory, and Notifications — plus NAF, which renders the avatar.
NAF stands for Native Avatar Framework: the engine that loads, animates, and renders Genies avatars. It stays separate from the modules because it has its own API and its own version number.
Imports
Auth, Inventory, and Notifications all arrive in one import. NAF is its own:
import GeniesPartnerSDK // Auth, Inventory, Notifications
import GeniesPartnerSDKNAF // avatar conveniences and chat
import NAFKit // rendering
Each module is a client you instantiate:
let auth = GeniesAuthClient()
let inventory = GeniesInventoryClient()
let notifications = GeniesNotificationsClient.shared
Adding one dependency does not mean one import — NAF is not re-exported through GeniesPartnerSDK, so you import it yourself.
A bridge between the two
GeniesPartnerSDKNAF is a second product in the same package, for the jobs that need both sides: rendering a user's avatar, dressing it, and talking to it.
import GeniesPartnerSDKNAF
Its convenience methods take Partner SDK types directly — hand them a GeniesAvatarRef or a wearable from the Inventory module and they do the right thing, which makes managing avatars and the link between the two much easier to follow. Reach past them to NAF when you want more direct control over loading, composition, or the render loop.
Chat lives here too, since a conversation is performed on the rendered avatar.
Where to look for what
| If you need to… | Module |
|---|---|
| Sign a user in, or refresh their session | Auth |
| See what a user owns | Inventory |
| List a user's avatars, or get a name, icon, or identifier | Inventory |
| React when ownership changes | Notifications |
| Render a user's avatar, dress it, or talk to it | GeniesPartnerSDKNAF |
| Drive loading, composition, or the render loop yourself | NAF |
The short version: the modules tell you which avatar; NAF draws it. The avatar APIs hand back identifiers and display data, not models or textures.
Avatars and smart avatars
There are two kinds of avatar, and the difference matters when you build a picker:
- An avatar is a look — body, face, and wearables. NAF renders it and can equip wearables on it.
- A smart avatar wraps an avatar with a personality, a voice, and a behavior. It renders identically, and it is the only kind that can hold a conversation.
Every avatar the SDK returns carries a canChat flag, true only for smart avatars. All three — personality, voice, behavior — are authored and published with the character, so what you choose from the SDK is which avatar to talk to, not how it talks. A plain avatar can't be upgraded to a smart avatar from the SDK either; that's a content-publishing step.
Handing an avatar to NAF
Avatar identifiers are opaque strings such as bm_smart-avatar_<hex>. You wrap whichever identifier you have in an avatar reference and pass it along — the SDK recognizes the format, so you never pick an endpoint or document type yourself.
That reference is the seam between the two: the Inventory module produces it, and NAF consumes it — most directly through view.loadAvatar(ref) or view.loadUserAvatar(ref) in GeniesPartnerSDKNAF.