Struct TurnContext 

pub struct TurnContext {

    pub session_id: u64,
    pub turn_number: u32,
    pub ttft_ms: u64,
    pub llm_duration_ms: u64,
    pub duration_ms: u64,
    pub tool_duration_ms: u64,
    pub usage: Option<UsageInfo>,
    pub full_text_len: u64,
    pub has_thinking: bool,
    pub tools_used: Vec<String>,
    pub tool_call_count: u32,
    pub tool_success: u32,
    pub tool_failed: u32,
    pub outcome: RunOutcome,
    pub error_message: Option<String>,
    pub user_input: String,
    pub model: String,
    pub plan_updates: u32,
    pub approval_count: u32,
    pub llm_calls: u32,
}

Pure-data snapshot of a completed turn iteration, passed to turn-end hooks.

Contains all raw data a consumer needs to build turn-level metrics — agent-base itself performs no aggregation, no persistence, and no business-level interpretation of this data.

Fields

session_id: u64

Numeric session identifier (agent-base internal).

turn_number: u32

1-based turn number within the session.

ttft_ms: u64

Time-to-first-token in milliseconds (user-perceived latency).

llm_duration_ms: u64

LLM stream duration in milliseconds.

duration_ms: u64

Total wall-clock turn duration in milliseconds (llm + tool + overhead).

tool_duration_ms: u64

Total tool execution duration in milliseconds.

usage: Option<UsageInfo>

Token usage from the LLM response, if available.

full_text_len: u64

Length of the full assistant text response in bytes.

has_thinking: bool

Whether the response included thinking/reasoning content.

tools_used: Vec<String>

Tool names called in this turn iteration.

tool_call_count: u32

Total number of tool calls made.

tool_success: u32

Number of tools that succeeded.

tool_failed: u32

Number of tools that failed.

outcome: RunOutcome

Outcome of this turn iteration (Completed / Failed / Cancelled / MaxTurnsExceeded).

error_message: Option<String>

Error message if the turn errored.

user_input: String

The user’s input text (may be truncated).

model: String

Model name used for the LLM call.

plan_updates: u32

Plan-update events emitted during this turn iteration (taken from EventBus).

approval_count: u32

Approval-request events emitted during this turn iteration (taken from EventBus).

llm_calls: u32

Number of LLM calls made (≥ 1; includes retries).

Trait Implementations

impl Clone for TurnContext

fn clone(&self) -> TurnContext

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for TurnContext

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

Formats the value using the given formatter.