Struct LlmClassifierRouteConfig 

pub struct LlmClassifierRouteConfig {

    pub classifier_target: String,
    pub mode: Option<ClassifierMode>,
    pub strong_target: Option<String>,
    pub weak_target: Option<String>,
    pub base_threshold: Option<f64>,
    pub threshold_step: Option<f64>,
    pub fail_open: Option<bool>,
    pub classify_trigger: ClassifyTrigger,
    pub message_hash_fallback: bool,
    pub recent_turn_window: Option<usize>,
    pub prompt: Option<String>,
    pub response_format_type: ClassifierResponseFormat,
    pub max_output_tokens: u64,
    pub escalation: Option<EscalationJudgeConfig>,
    pub decision: Option<DecisionJudgeRouteConfig>,
    pub models: Option<CategoryModelConfig>,
    pub default_target: Option<String>,
    pub response_schema: Option<String>,
    pub policy: Option<ClassifierPolicyConfig>,
}

Settings for an llm_classifier route. Which fields are required depends on the ClassifierMode; using a field from the wrong mode is an error.

Fields

classifier_target: String

Target the judge is called through. Never a routing destination itself.

mode: Option<ClassifierMode>

Mode to run. Defaults to escalation when escalation is set, otherwise capability.

strong_target: Option<String>

Capability and escalation modes: the capable tier.

weak_target: Option<String>

Capability and escalation modes: the efficient tier.

base_threshold: Option<f64>

Capability mode: lowest solve probability that still routes to the weak target, from 0 to 1.

threshold_step: Option<f64>

Capability mode: how much to raise the threshold when the judge is uncertain. Added once for an uncertain verdict and twice for unsupported.

fail_open: Option<bool>

Capability mode: routes to the capable tier on judge client failures and deadlines. Defaults to true when capability mode is selected.

classify_trigger: ClassifyTrigger

How often the judge runs: every request, once per user turn, or once per session.

message_hash_fallback: bool

Reuses the session’s target by hashing the first user message when no session ID is available. Needs classify_trigger = "new_session".

recent_turn_window: Option<usize>

How many trailing turns the judge sees. Unset shows it the opening task and the latest user follow-up only.

prompt: Option<String>

Replaces the packaged judge prompt. Required in custom mode.

response_format_type: ClassifierResponseFormat

How the judge is asked for structured output. Use json_object when the provider cannot do JSON Schema.

max_output_tokens: u64

Most completion tokens the judge verdict may use.

escalation: Option<EscalationJudgeConfig>

Escalation mode: how many escalate verdicts latch the session, and how much of the transcript the judge sees.

decision: Option<DecisionJudgeRouteConfig>

Capability mode: use a decision target as the judge, with relative-advantage scoring.

models: Option<CategoryModelConfig>

Custom mode: runtime model groups.

default_target: Option<String>

Custom mode: category used when the judge fails or its verdict cannot be routed.

response_schema: Option<String>

Custom mode: JSON Schema the verdict must match, written as a string.

policy: Option<ClassifierPolicyConfig>

Custom mode: how to read the chosen target out of the verdict.

Trait Implementations

impl Clone for LlmClassifierRouteConfig

fn clone(&self) -> LlmClassifierRouteConfig

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for LlmClassifierRouteConfig

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

Formats the value using the given formatter.

impl Default for LlmClassifierRouteConfig

fn default() -> LlmClassifierRouteConfig

Returns the “default value” for a type.

impl<'de> Deserialize<'de> for LlmClassifierRouteConfig

where LlmClassifierRouteConfig: Default,

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

where __D: Deserializer<'de>,

Deserialize this value from the given Serde deserializer.