Streaming

The runtime supports streamed chat output through ChatChunk and ChatExecutionEvent.

Chat streams

Each stream / stream_prompt / stream_* call on Chat performs one model run. Events are deltas plus a final ChatChunk::Finished for that turn.

Agent streams

Streaming on Agent uses the same chunk types, but the stream may include multiple model rounds while tools are requested. Between rounds you may see ChatExecutionEvent::ToolCall and ToolResult events after a finished assistant message that contained tool calls.

JavaScript note

streamPrompt / streamPromptWithContent collect the full event list in order (one round-trip through the WASM bridge); a browser ReadableStream API is not wired yet.

Event coverage

Streaming currently includes message start events, content deltas, tool call deltas, and final response events (per model turn; agent streams may emit several before the stream ends).

See the README “Streaming” section for additional design notes.