Getting Started

Installation

For a complete local setup with presets:

npm install @flowrag/pipeline @flowrag/presets

Or install individual packages:

npm install @flowrag/core @flowrag/pipeline \
  @flowrag/storage-json @flowrag/storage-sqlite @flowrag/storage-lancedb \
  @flowrag/provider-local @flowrag/provider-gemini

For AWS cloud deployment:

npm install @flowrag/provider-bedrock @flowrag/storage-s3 @flowrag/storage-opensearch

Quick Start

import { defineSchema } from '@flowrag/core';
import { createFlowRAG } from '@flowrag/pipeline';
import { createLocalStorage } from '@flowrag/presets';

// 1. Define your schema
const schema = defineSchema({
  entityTypes: ['SERVICE', 'DATABASE', 'PROTOCOL'],
  relationTypes: ['USES', 'PRODUCES', 'CONSUMES'],
});

// 2. Create RAG instance
const rag = createFlowRAG({
  schema,
  ...createLocalStorage('./data'),
});

// 3. Index documents
await rag.index('./content');

// 4. Search
const results = await rag.search('how does authentication work');

That's it. FlowRAG scans your files, chunks them, extracts entities and relations via LLM, generates embeddings, and stores everything locally.

What Happens Under the Hood

When you call rag.index():

1. Scanner reads files from the input path
2. Chunker splits documents into ~1200 token chunks with overlap
3. Extractor (LLM) identifies entities and relations in each chunk
4. Embedder generates vector embeddings for semantic search
5. Storage saves documents, vectors, and the knowledge graph

When you call rag.search():

1. Your query is embedded into a vector
2. Vector search finds semantically similar chunks
3. Graph traversal follows entity relationships for connected context
4. Results are merged, deduplicated, and optionally reranked

What's Next?

• Schema Definition — custom entity types, relation types, and fields
• Indexing — chunking, caching, incremental indexing
• Querying — query modes, dual retrieval, data flow tracing
• Providers — choose your embedder, extractor, and reranker
• Deployment — local CLI or AWS Lambda