@flowrag/core

Interfaces, schema definition, and types for FlowRAG.

Installation

npm install @flowrag/core

Usage

Schema Definition

import { defineSchema } from '@flowrag/core';

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

schema.isValidEntityType('SERVICE');     // true
schema.normalizeEntityType('UNKNOWN');   // 'Other'

Custom Fields

const schema = defineSchema({
  entityTypes: ['SERVICE', 'DATABASE'],
  relationTypes: ['USES', 'PRODUCES'],
  entityFields: {
    status: { type: 'enum', values: ['active', 'deprecated'], default: 'active' },
  },
  relationFields: {
    syncType: { type: 'enum', values: ['sync', 'async'] },
  },
});

Interfaces

• KVStorage — key-value storage (documents, chunks, cache)
• VectorStorage — vector embeddings for semantic search
• GraphStorage — knowledge graph (entities, relations, traversal)
• Embedder — text to vector embeddings
• LLMExtractor — entity/relation extraction via LLM
• Reranker — post-retrieval result reranking

License

MIT