Search and filters

import { vectorFilter } from '@anvia/core/vector-store'

const results = await store.index(embeddings).search({
  query: 'reset a password',
  topK: 10,
  filter: vectorFilter.or(
    vectorFilter.eq('tier', 'public'),
    vectorFilter.gt('priority', 3),
  ),
})

The adapter translates eq, gt, lt, and, and or to a Lance SQL expression. It escapes string literals, renders booleans as TRUE or FALSE, and uses IS NULL for equality with null.

Filter keys become identifiers in the expression, so use application-defined metadata names rather than arbitrary user input. Retrieval filters do not replace authorization.

Search scores are 1 - distance and are meaningful only for the selected distance and corpus.