OptionalcacheOptionalquery_OptionalrerankingOptionalenabled?: booleanOptionalmatch_threshold?: numberMatch threshold (0-1, default 0.4)
Optionalmodel?: stringOptionalretrievalOptionalboost_by?: { direction?: "asc" | "desc" | "exists" | "not_exists"; field: string }[]Boost results by metadata field values. Max 3 entries.
Optionalcontext_expansion?: numberNumber of surrounding chunks to include for context (0-3). Default 0.
Optionalfilters?: VectorizeVectorMetadataFilterVectorize metadata filters applied to the search.
Optionalfusion_method?: "max" | "rrf"Fusion method for combining vector + keyword results.
Optionalkeyword_match_mode?: "and" | "or"How keyword terms are combined: "and" = all terms must match, "or" = any term matches.
Optionalmatch_threshold?: numberMinimum similarity score (0-1) for a result to be included. Default 0.4.
Optionalmax_num_results?: numberMaximum number of results to return (1-50). Default 10.
Optionalmetadata_only?: booleanIf true, return only item metadata without chunk text.
Optionalretrieval_type?: "vector" | "keyword" | "hybrid"Which retrieval backend to use. Defaults to the instance's configured index_method.
Optionalreturn_on_failure?: booleanIf true (default), return empty results on retrieval failure instead of throwing.
Common shape for
ai_search_optionsused by both single-instance and multi-instance requests. Contains retrieval, query rewrite, reranking, and cache sub-options.