streaming

Streaming protocol implementation guide for custom AI backends and assistant-ui integration. Always consult assistant-ui.com/llms.txt for latest API. The assistant-stream package handles streaming from AI backends. useLocalRuntime expects ChatModelRunResult chunks. Yield content parts for streaming: Stream not updating UI Tool calls not rendering Partial text not showing - Provides createAssistantStreamResponse for building custom streaming responses with text, tool calls, and structured event handling - Supports two integration paths: Vercel AI SDK (via toUIMessageStreamResponse ) or native assistant-stream for non-SDK backends - Includes useLocalRuntime pattern for client-side streaming with ChatModelRunResult chunks and content part yielding - Covers stream event types (part-start, text-delta, result, step-finish, error) and debugging via AssistantStream and DataStreamDecoder - Documents common issues: Content-Type headers, CORS errors, tool call registration, and text-delta event requirements - ./references/data-stream.md -- AI SDK data stream format - ./references/assistant-transport.md -- Native assistant-ui format - ./references/encoders.md -- Encoders and decoders - part-start with part.type = "text" | "reasoning" | "tool-call" | "source" | "file" - text-delta with streamed text - result with tool results - step-start , step-finish , message-finish - error strings -...