Skip to content

Redis Storage

@flowrag/storage-redis provides KV and vector storage on Redis Stack, combining fast key-value operations with vector similarity search via RediSearch.

Terminal window
npm install @flowrag/storage-redis redis
import { createClient } from 'redis';
import { RedisKVStorage, RedisVectorStorage } from '@flowrag/storage-redis';
const client = createClient({ url: 'redis://localhost:6379' });
await client.connect();
const kv = new RedisKVStorage({ client });
const vector = new RedisVectorStorage({ client, dimensions: 384 });

Key-value storage for documents, chunks, and LLM cache.

const kv = new RedisKVStorage({
client,
prefix: 'flowrag:kv:', // default
});
Option Type Default Description
client RedisClient — Connected redis client (required)
prefix string flowrag:kv: Key prefix for namespacing

Vector similarity search using RediSearch (HNSW index, cosine distance).

const vector = new RedisVectorStorage({
client,
dimensions: 384,
index: 'flowrag-vectors', // default
prefix: 'flowrag:vec:', // default
});
Option Type Default Description
client RedisClient — Connected redis client (required)
dimensions number — Vector dimensions (required)
index string flowrag-vectors RediSearch index name
prefix string flowrag:vec: Hash key prefix

The index is created automatically on first use with HNSW algorithm and COSINE distance metric.

import { defineSchema } from '@flowrag/core';
import { createFlowRAG } from '@flowrag/pipeline';
import { createClient } from 'redis';
import { RedisKVStorage, RedisVectorStorage } from '@flowrag/storage-redis';
import { SQLiteGraphStorage } from '@flowrag/storage-sqlite';
import { LocalEmbedder } from '@flowrag/provider-local';
import { GeminiExtractor } from '@flowrag/provider-gemini';
const client = createClient({ url: 'redis://localhost:6379' });
await client.connect();
const rag = createFlowRAG({
schema: defineSchema({
entityTypes: ['SERVICE', 'DATABASE'],
relationTypes: ['USES', 'PRODUCES'],
}),
storage: {
kv: new RedisKVStorage({ client }),
vector: new RedisVectorStorage({ client, dimensions: 384 }),
graph: new SQLiteGraphStorage({ path: './data/graph.db' }),
},
embedder: new LocalEmbedder(),
extractor: new GeminiExtractor(),
});