Struct LlmClassifier 

pub struct LlmClassifier {
    pub model: Option<String>,
}

LLM-based classifier for refining ambiguous heuristic scores.

When classifier_model is configured (e.g. "anthropic/claude-haiku-4"), this classifier uses a fast/cheap model to classify ambiguous requests. It only activates when the heuristic score falls in the ambiguous zone (0.25–0.75).

The LLM understands any language natively — no keyword lists needed.

Fields

model: Option<String>

The model to use for classification, in "provider/model-id" format.

Implementations

impl LlmClassifier

pub fn new(model: Option<String>) -> Self

Create a new LLM classifier.

pub async fn classify( &self, input: &ClassifierInput, heuristic_score: f64, ) -> Result<f64>

Classify a user message to determine routing tier.

Sends a minimal prompt to the configured classifier model and parses the response as “high”, “medium”, or “low”.

Returns a score:

• 0.1 for low
• 0.5 for medium
• 0.9 for high

Falls back to the provided heuristic_score on any error.

Trait Implementations

impl Clone for LlmClassifier

fn clone(&self) -> LlmClassifier

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for LlmClassifier

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

Formats the value using the given formatter.

impl Default for LlmClassifier

fn default() -> LlmClassifier

Returns the “default value” for a type.