Struct SemanticRecallParams 

pub struct SemanticRecallParams<'a> {
    pub query: &'a str,
    pub token_budget: usize,
    pub recall_limit: usize,
    pub context_format: ContextFormat,
    pub conversation_id: Option<ConversationId>,
    pub tiered_classifier: Option<&'a Arc<AnyProvider>>,
    pub tiered_validator: Option<&'a Arc<AnyProvider>>,
    pub tiered_config: &'a TieredRetrievalConfig,
}

Configuration parameters for semantic recall injection.

Collects the 8 config-like arguments shared between the tiered and flat recall paths so callers do not need to pass them positionally to ContextService::inject_semantic_recall_bare.

window and memory are kept as direct parameters on the method because they are mutable/output args rather than configuration.

Fields

query: &'a str

Query string used for retrieval.

token_budget: usize

Maximum number of tokens the injected recall may consume.

recall_limit: usize

Maximum number of memories to retrieve (flat path only).

context_format: ContextFormat

Format applied when serialising recalled memories.

conversation_id: Option<ConversationId>

Conversation scope used for tiered retrieval.

tiered_classifier: Option<&'a Arc<AnyProvider>>

Optional LLM provider for intent classification (tiered path).

tiered_validator: Option<&'a Arc<AnyProvider>>

Optional LLM provider for result validation (tiered path).

tiered_config: &'a TieredRetrievalConfig

Tiered retrieval configuration controlling whether the tiered path is active.