Group: Ynagai Agui

Sort by:Popular▼

✓Popular✓Newest

Draws an agui-model transcript with Compose runtime and foundation only: no design system, no Markdown parser, one overridable slot per part kind.

The render model for an AG-UI client: ordered message parts with per-part streaming state, independent of any UI framework.

Reads A2UI out of an AG-UI transcript -- activities, streamed render calls and tool results -- and turns upstream's v0.9 envelopes into the v1.0 messages a2ui-compose parses. No Compose.

Keeps an a2ui-compose renderer up to date with an AG-UI transcript and fills agui-compose's activity and tool-call slots with its surfaces. No design system.

Folds an AG-UI event stream into the agui-model render model, including the ACTIVITY_SNAPSHOT and ACTIVITY_DELTA events the upstream reducer does not handle.

Fills every slot of agui-compose's renderer table with Material 3: role-framed messages, a reasoning disclosure, and tool-call, activity and attachment surfaces.

The A2UI basic catalog's Material 3 renderers and a Material 3 building-UI state, so a Material 3 AG-UI transcript draws A2UI with one call.

Runs an upstream AG-UI agent and keeps its transcript: one render model per thread, fed by every run, observable as a StateFlow.

Draws agent prose as Markdown through agui-compose's text renderer slot, parsing incrementally while a response is still streaming.