Struct SessionAutoModeResolvedData 

pub struct SessionAutoModeResolvedData {

    pub available_models: Option<Vec<String>>,
    pub candidate_models: Option<Vec<String>>,
    pub category_scores: Option<HashMap<String, f64>>,
    pub chosen_model: String,
    pub chosen_shortfall: Option<f64>,
    pub confidence: Option<f64>,
    pub end_to_end_latency_ms: Option<f64>,
    pub fallback: Option<bool>,
    pub fallback_reason: Option<String>,
    pub has_image: Option<bool>,
    pub predicted_label: Option<String>,
    pub reasoning_bucket: Option<AutoModeResolvedReasoningBucket>,
    pub router_latency_ms: Option<f64>,
    pub routing_method: Option<String>,
    pub sticky_override: Option<bool>,
}

Session event “session.auto_mode_resolved”. Auto Intent resolution: the concrete model the session settled on for the first prompt of an auto-mode session, and why. Lets SDK clients render the chosen model and the full reason it was picked. The core selection fields (chosenModel/reasoningBucket/categoryScores) are stable; the routing-analytics fields (predictedLabel/confidence/candidateModels) mirror the upstream intent service and may evolve, hence the event’s experimental stability.

Experimental. This type is part of an experimental wire-protocol surface and may change or be removed in future SDK or CLI releases.

Fields

available_models: Option<Vec<String>>

Models offered to the router for this resolution

candidate_models: Option<Vec<String>>

Ordered candidate model list the router returned, when not a fallback

category_scores: Option<HashMap<String, f64>>

Per-category classifier scores (0-1) behind the bucket: the granular HYDRA capability scores (reasoning, code_gen, debugging, tool_use), or the binary needs_reasoning/no_reasoning scores when HYDRA didn’t run. Lets clients show a breakdown rather than just the bucket.

chosen_model: String

The concrete model the session will use after any intent refinement

chosen_shortfall: Option<f64>

The chosen model’s score shortfall relative to the top candidate

confidence: Option<f64>

Classifier confidence for the predicted label, when available

end_to_end_latency_ms: Option<f64>

End-to-end client wait time for the router request in milliseconds

fallback: Option<bool>

Whether the router fell back to the standard Auto selection

fallback_reason: Option<String>

Server-provided reason for falling back, when available

has_image: Option<bool>

Whether the routed prompt contained an image

predicted_label: Option<String>

The predicted classifier label (e.g. needs_reasoning), when available

reasoning_bucket: Option<AutoModeResolvedReasoningBucket>

Coarse request-difficulty bucket, for explaining why a model was chosen (“picked X because this looks like high-reasoning work”)

router_latency_ms: Option<f64>

Server-reported router processing time in milliseconds

routing_method: Option<String>

The routing method the server applied, when Auto Intent ran

sticky_override: Option<bool>

Whether a sticky model choice overrode the router result

Trait Implementations

impl Clone for SessionAutoModeResolvedData

fn clone(&self) -> SessionAutoModeResolvedData

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for SessionAutoModeResolvedData

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

Formats the value using the given formatter.

impl Default for SessionAutoModeResolvedData

fn default() -> SessionAutoModeResolvedData

Returns the “default value” for a type.

impl<'de> Deserialize<'de> for SessionAutoModeResolvedData

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

where __D: Deserializer<'de>,

Deserialize this value from the given Serde deserializer.

impl Serialize for SessionAutoModeResolvedData

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

where __S: Serializer,

Serialize this value into the given Serde serializer.