Storage Backends

Mnemo supports 4 storage backends. LanceDB is the default — embedded, zero-config, no external services needed.

LanceDB (Default)

Embedded vector database. No setup required.

const mnemo = await createMnemo({
  embedding: { /* ... */ },
  dbPath: './my-memory-db',
  // storageBackend: 'lancedb',  // default, can be omitted
});

Pros: Zero config, embedded, fast, full BM25 support Cons: Single-process access (no concurrent writers)

Qdrant

High-performance vector database with filtering.

# Start Qdrant
docker run -p 6333:6333 qdrant/qdrant

# Install driver
npm install @qdrant/js-client-rest

const mnemo = await createMnemo({
  embedding: { /* ... */ },
  dbPath: './qdrant-memories',
  storageBackend: 'qdrant',
  storageConfig: {
    url: 'http://localhost:6333',
    collectionName: 'memories',    // optional, default: 'mnemo_memories'
  },
});

Pros: Production-grade, concurrent access, filtering, cloud-hosted option Cons: Requires running a server, no native BM25

Chroma

Open-source embedding database.

# Start Chroma
docker run -p 8000:8000 chromadb/chroma

# Install driver
npm install chromadb

const mnemo = await createMnemo({
  embedding: { /* ... */ },
  dbPath: './chroma-memories',
  storageBackend: 'chroma',
  storageConfig: {
    url: 'http://localhost:8000',
    collectionName: 'memories',
  },
});

Pros: Simple, Python ecosystem integration Cons: Requires server, no native BM25

PGVector

PostgreSQL with vector extensions — use your existing database.

# Install driver
npm install pg

# Enable pgvector in your PostgreSQL
CREATE EXTENSION vector;

const mnemo = await createMnemo({
  embedding: { /* ... */ },
  dbPath: './pg-memories',
  storageBackend: 'pgvector',
  storageConfig: {
    connectionString: 'postgresql://user:pass@localhost:5432/mydb',
    tableName: 'memories',         // optional, default: 'mnemo_memories'
  },
});

Pros: Use existing Postgres, SQL queries, battle-tested Cons: Requires PostgreSQL + pgvector extension

Comparison

Backend	Setup	BM25	Concurrent	Cloud Option
LanceDB	Zero config	✅	Single writer	LanceDB Cloud
Qdrant	Docker/cloud	❌	✅	Qdrant Cloud
Chroma	Docker	❌	✅	—
PGVector	PostgreSQL	❌	✅	Any managed PG

Custom Adapters

You can register your own storage adapter:

import { registerAdapter, type StorageAdapter } from '@mnemoai/core/storage-adapter';

registerAdapter('my-backend', (config) => {
  return {
    // implement StorageAdapter interface
  } as StorageAdapter;
});