memori.ts

Quickstart

Build a memory-augmented agent in 30 seconds.

1. Initialization

Import Memori and your LLM client. Memori will default to using a local SQLite database in your project root.

typescript

1import { Memori } from "memori-js";
2import OpenAI from "openai";
3
4const memori = new Memori(); 
5const client = new OpenAI({ apiKey: process.env.OPENAI_API_KEY });

2. Registration

Connect the two clients. This registers a middleware that intercepts calls to chat.completions.create.

typescript

1// This single line enables active memory
2memori.llm.register(client);

3. Interaction

Chat normally. Memori silently searches for relevant past conversations and injects them into the system prompt.

typescript

1async function chat(message: string) {
2  const response = await client.chat.completions.create({
3    model: "gpt-4",
4    messages: [
5      { role: "user", content: message }
6    ]
7  });
8  
9  console.log(response.choices[0].message.content);
10}
11
12// First run: "My name is Roach." -> Saved to memory.
13// Second run: "What is my name?" -> Retrieves context, answers "Roach".

Explore Core Concepts