@lineai/memory

A functional TypeScript client library for the Cognee API. This library provides a simple, type-safe interface for interacting with Cognee's knowledge graph and memory management services.

Installation

npm install @lineai/cognee-api
# or
yarn add @lineai/cognee-api

Quick Start

import { CogneeConfig, login, addData, cognify, search, SearchType } from '@lineai/cognee-api';

// Configure the client
const config: CogneeConfig = {
  baseUrl: 'http://localhost:8000',
};

// Login (if authentication is required)
await login(config, {
  username: 'your-username',
  password: 'your-password',
});

// Add data to a dataset
const file = new File(['content'], 'document.txt');
await addData(config, [file], {
  datasetName: 'my-dataset',
});

// Process the data into a knowledge graph
await cognify(config, {
  datasets: ['my-dataset'],
});

// Search the knowledge graph
const results = await search(config, {
  query: 'What is in the document?',
  dataset_name: 'my-dataset',
  search_type: SearchType.GRAPH_COMPLETION,
});

console.log(results);

Core Concepts

Configuration

All API functions require a CogneeConfig object:

interface CogneeConfig {
  baseUrl: string;  // The Cognee API base URL
  apiKey?: string;  // Optional API key for cloud services
}

Functional API

All functions follow a simple pattern:

(config: CogneeConfig, ...params) => Promise<Result>

This makes them easy to compose, test, and use with different configurations.

API Reference

See the full documentation below for all available functions and types.

Philosophy

This library follows functional programming principles:

• Pure functions: All API calls are pure functions of their inputs
• Immutability: All data structures use readonly properties
• Composition: Functions are designed to be easily composed
• Minimal abstraction: Direct mapping to API endpoints
• Type safety: Full TypeScript support

License

MIT