Struct LlmStreamEvent 

pub struct LlmStreamEvent {
    pub call_id: String,
    pub provider: String,
    pub model: String,
    pub delta: String,
    pub done: bool,
    pub error: Option<String>,
}

A single chunk of generated text emitted as the model produces tokens.

delta is the new text since the previous event; the consumer is responsible for accumulating if it needs the full response. done is set exactly once per call - either with the final partial delta or on its own with an empty delta - and signals that the stream is closed.

error is non-empty when the call failed mid-stream; done will also be true in that case. Successful calls leave error None.

Fields

call_id: String

Caller-supplied id correlating this event to a specific call. Lets the frontend group deltas when multiple LLM calls overlap.

provider: String

Provider name (e.g. "local", "claude").

model: String

Model id the provider reported (may differ from the requested id when the server aliases). Empty until the first chunk arrives.

delta: String

New text since the previous event.

done: bool

True on the final event (either with the last delta or empty).

error: Option<String>

Set when the call failed mid-stream; the consumer should surface this to the user (e.g. an inline error tag on the chip).

Implementations

impl LlmStreamEvent

pub fn final_delta(call_id: &str, resp: &CloudAiResponse) -> Self

Construct the single done event the default invoke_stream impl emits when wrapping a non-streaming invoke call.

pub fn error(call_id: &str, provider: &str, error: String) -> Self

Construct a terminal error event so consumers know the stream is over.

Trait Implementations

impl Clone for LlmStreamEvent

fn clone(&self) -> LlmStreamEvent

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for LlmStreamEvent

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.