pub struct MemorySearch {Show 17 fields
pub query: Option<String>,
pub embedding: Option<Vec<f32>>,
pub limit: usize,
pub namespace: Option<String>,
pub session_id: Option<String>,
pub tags: Vec<String>,
pub embedding_model: Option<String>,
pub candidate_limit: usize,
pub fusion: MemoryFusion,
pub rrf_k: f32,
pub vector_weight: f32,
pub lexical_weight: f32,
pub recency_weight: f32,
pub importance_weight: f32,
pub half_life_seconds: u64,
pub min_score: f32,
pub exact: bool,
}Expand description
Search text, supplied embeddings, or both, with explicit source filters.
Fields§
§query: Option<String>BM25 query text. At least query or embedding is required.
embedding: Option<Vec<f32>>An embedding from the same model/dimension as stored documents.
limit: usizeMaximum results, from zero to 10,000.
namespace: Option<String>Optional namespace restriction.
session_id: Option<String>Optional session restriction.
Match at least one requested tag when this list is nonempty.
embedding_model: Option<String>Embedding model identity; vector searches never mix model spaces.
candidate_limit: usizeTop candidates per channel before final ranking.
fusion: MemoryFusionRank fusion or bounded weighted scoring.
rrf_k: f32RRF smoothing constant.
vector_weight: f32Semantic contribution weight.
lexical_weight: f32Keyword contribution weight.
recency_weight: f32Recency contribution weight.
importance_weight: f32Importance contribution weight.
half_life_seconds: u64Recency half-life in seconds; zero disables decay.
min_score: f32Minimum final ranking score, from zero to one.
exact: boolDeliberate exact vector scan for evaluation/small corpora.