Anvia

Appearance

@anvia/lancedb ​

@anvia/lancedb stores Anvia embedded documents in LanceDB. It suits local-first retrieval, development, and deployments that already own a LanceDB connection.

Install ​

sh
pnpm add @anvia/lancedb @anvia/core @anvia/openai @lancedb/lancedb

The ESM package includes @lancedb/lancedb and peers with @anvia/core >=0.7.1 <1.0.0.

Store and search documents ​

ts
import { embedDocuments } from '@anvia/core/embeddings'
import { LanceDBVectorStore } from '@anvia/lancedb'
import { OpenAIClient } from '@anvia/openai'

const openai = new OpenAIClient({
  apiKey: process.env.OPENAI_API_KEY,
})
const embeddings = openai.embeddingModel('text-embedding-3-small')
const sourceDocuments = [
  {
    id: 'password-reset',
    text: 'Password reset links expire after 30 minutes.',
    category: 'account',
  },
]

const documents = await embedDocuments(embeddings, sourceDocuments, {
  id: (document) => document.id,
  content: (document) => document.text,
  metadata: (document) => ({ category: document.category }),
})

const store = await LanceDBVectorStore.connect({
  uri: 'data/lancedb',
  tableName: 'support_docs',
  vectorSize: 1536,
})

await store.upsertDocuments(documents)

const results = await store.index(embeddings).search({
  query: 'How do I reset a password?',
  topK: 5,
})

If neither client nor uri is supplied, the adapter connects to ~/.anvia/lancedb.

Table ownership ​

connect() creates an empty table when it is missing unless createIfMissing: false. Production applications should supply a managed connection and own table lifecycle, indexing, backups, and optimization through their LanceDB deployment process.

The adapter writes reserved __anvia_ columns for the logical document ID, serialized document, and vector. Metadata keys beginning with __anvia_ are rejected. Repeated ingestion adds rows through LanceDB's add() API; plan deduplication and cleanup around your ingestion workflow.

Production patterns ​

  • Use an explicit durable URI or injected connection; do not depend on a home-directory default in containers.
  • Keep vectorSize aligned with the embedding model.
  • Provision and tune indexes outside request handling for larger datasets.
  • Use stable source IDs and an intentional replace/delete strategy when refreshing a corpus.
  • Monitor table growth when documents produce more than one embedding.

Learn the common workflow in Load documents and Vector stores.

Reference ​

Built for Anvia.