Enum StepEvent 

#[non_exhaustive]
pub enum StepEvent {
    AssistantText {
        text: String,
        step: usize,
    },
    ToolCall {
        call: ToolCall,
        step: usize,
    },
    Latency {
        step: usize,
        llm_ms: u64,
    },
    ToolResult {
        id: String,
        name: String,
        output: String,
        step: usize,
    },
    Usage {
        usage: TokenUsage,
        step: usize,
    },
    PartialToken {
        text: String,
        step: usize,
    },
    Compacted {
        removed: usize,
        kept: usize,
        summary_chars: usize,
        step: usize,
    },
    Finished {
        reason: FinishReason,
        steps: usize,
    },
    PlanProposed {
        plan_text: String,
        tool_calls: Vec<ToolCall>,
    },
    PlanConfirmed,
    PlanRejected {
        reason: String,
    },
}

Variants (Non-exhaustive)

Non-exhaustive enums could have additional variants added in future. Therefore, when matching against variants of non-exhaustive enums, an extra wildcard arm must be added to account for any future variants.

AssistantText

Model generated text without tool calls.

Emitted when the LLM produces a response. This is typically the final answer or intermediate reasoning. The text field contains the complete model response, and step indicates which iteration this occurred on.

Fields

text: String

step: usize

ToolCall

Model requested to execute a tool.

Emitted when the LLM calls a tool. Contains the tool name, ID, and arguments that will be executed. The call is dispatched to the registry after this event. Tool errors are reported via ToolResult events, not ToolCall failures.

Fields

call: ToolCall

step: usize

Latency

Time taken for the LLM request (excluding tool execution).

Emitted after the model responds. Useful for measuring provider latency and diagnosing slow responses. llm_ms is in milliseconds.

Fields

step: usize

llm_ms: u64

ToolResult

Result of executing a tool call.

Emitted after a tool finishes executing. Contains the tool name, call ID, the output string (or error message), and the step number. This result is added to the transcript and sent back to the model for the next iteration. In case of tool error, the output will be the error message prefixed with “ERROR: “.

Fields

id: String

name: String

output: String

step: usize

Usage

Token usage statistics from the LLM provider.

Emitted if the provider returns usage information (input tokens, output tokens). Accumulated across all steps for total usage. Useful for cost tracking and monitoring resource consumption.

Fields

usage: TokenUsage

step: usize

PartialToken

Partial token from streaming response (if streaming enabled).

Only emitted if streaming is enabled. Contains a single token or partial chunk of the model’s response text. Allows UI layers to display real-time incremental updates to the model’s output without waiting for the entire response.

Fields

text: String

step: usize

Compacted

Transcript was compacted to fit size constraints.

Emitted when the transcript exceeds the max size and is automatically compacted. The removed field shows how many messages were summarized, kept shows how many remain, and summary_chars shows the size of the compaction summary added. This allows UI layers to notify users that older context has been summarized. Compaction only occurs if a Compactor is configured on the Agent.

Fields

removed: usize

kept: usize

summary_chars: usize

step: usize

Finished

Agent run completed.

Emitted as the final event. Indicates the run is done and why it stopped. reason explains the termination (no more tool calls, budget exceeded, stuck detection, etc.). steps shows how many iterations were executed. After this event, no more events will be emitted for this run.

Fields

reason: FinishReason

steps: usize

PlanProposed

Agent has produced a plan and is waiting for confirmation.

Fields

plan_text: String

Human-readable plan description

tool_calls: Vec<ToolCall>

The buffered tool calls

PlanConfirmed

Plan was confirmed, execution will proceed.

PlanRejected

Plan was rejected with a reason.

Fields

reason: String

Trait Implementations

impl Clone for StepEvent

fn clone(&self) -> StepEvent

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for StepEvent

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

Formats the value using the given formatter.

impl<'de> Deserialize<'de> for StepEvent

fn deserialize<__D>(__deserializer: __D) -> Result<Self, __D::Error>

where __D: Deserializer<'de>,

Deserialize this value from the given Serde deserializer.

impl Serialize for StepEvent

fn serialize<__S>(&self, __serializer: __S) -> Result<__S::Ok, __S::Error>

where __S: Serializer,

Serialize this value into the given Serde serializer.