FluxbaseVector

FluxbaseVector provides vector search functionality using pgvector

Example

// Embed text and search
const { data: results } = await client.vector.search({
  table: 'documents',
  column: 'embedding',
  query: 'How to use TypeScript?',
  match_count: 10
})

// Embed text directly
const { data: embedding } = await client.vector.embed({ text: 'Hello world' })

Constructors

new FluxbaseVector()

new FluxbaseVector(fetch): FluxbaseVector

Parameters

Parameter	Type
fetch	FluxbaseFetch

Returns

FluxbaseVector

Methods

embed()

embed(request): Promise<FluxbaseResponse<EmbedResponse>>

Generate embeddings for text

Parameters

Parameter	Type
request	EmbedRequest

Returns

Promise<FluxbaseResponse<EmbedResponse>>

Example

// Single text
const { data } = await client.vector.embed({
  text: 'Hello world'
})
console.log(data.embeddings[0]) // [0.1, 0.2, ...]

// Multiple texts
const { data } = await client.vector.embed({
  texts: ['Hello', 'World'],
  model: 'text-embedding-3-small'
})

---

search()

search<T>(options): Promise<FluxbaseResponse<VectorSearchResult<T>>>

Search for similar vectors with automatic text embedding

This is a convenience method that:

1. Embeds the query text automatically (if query is provided)
2. Performs vector similarity search
3. Returns results with distance scores

Type Parameters

Type Parameter	Default type
T	Record<string, unknown>

Parameters

Parameter	Type
options	VectorSearchOptions

Returns

Promise<FluxbaseResponse<VectorSearchResult<T>>>

Example

// Search with text query (auto-embedded)
const { data } = await client.vector.search({
  table: 'documents',
  column: 'embedding',
  query: 'How to use TypeScript?',
  match_count: 10,
  match_threshold: 0.8
})

// Search with pre-computed vector
const { data } = await client.vector.search({
  table: 'documents',
  column: 'embedding',
  vector: [0.1, 0.2, ...],
  metric: 'cosine',
  match_count: 10
})

// With additional filters
const { data } = await client.vector.search({
  table: 'documents',
  column: 'embedding',
  query: 'TypeScript tutorial',
  filters: [
    { column: 'status', operator: 'eq', value: 'published' }
  ],
  match_count: 10
})