@flowrag/provider-gemini

Gemini AI provider for FlowRAG - embeddings and entity extraction.

Installation

npm install @flowrag/provider-gemini

Usage

Embedder

import { GeminiEmbedder } from '@flowrag/provider-gemini';

const embedder = new GeminiEmbedder({
  apiKey: 'your-gemini-api-key', // or set GEMINI_API_KEY env var
  model: 'text-embedding-004', // optional, default
});

// Single embedding
const embedding = await embedder.embed('Hello world');

// Batch embeddings
const embeddings = await embedder.embedBatch(['Hello', 'World']);

Extractor

import { GeminiExtractor } from '@flowrag/provider-gemini';
import { defineSchema } from '@flowrag/core';

const extractor = new GeminiExtractor({
  apiKey: 'your-gemini-api-key', // or set GEMINI_API_KEY env var
  model: 'gemini-2.0-flash-exp', // optional, default
  temperature: 0.1, // optional, default
});

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

const result = await extractor.extractEntities(
  'ServiceA connects to DatabaseB',
  ['ServiceC'], // known entities
  schema
);

Environment Variables

GEMINI_API_KEY=your-api-key

License

MIT