Base AI

The AI framework for building declarative and composable AI-powered LLM products.

Getting Started with @baseai/core

Installation

First, install the @baseai/core package using npm or yarn:

npm install @baseai/core

or

pnpm add @baseai/core

or

yarn add @baseai/core

Usage

To use the generate function from the @baseai/core package, follow these steps:

1. Import the generate function:

   import {generate} from '@baseai/core';

2. Set up environment variables:

   Ensure you have the following environment variables set in your .env file:

   OPENAI_API_KEY=your_openai_api_key

3. Generate a response using a prompt:

   import {generate} from '@baseai/core';
   
   async function exampleWithPrompt() {
   	const response = await generate({
   		model: 'gpt-3.5-turbo-0125',
   		prompt: '1+1',
   	});
   
   	console.log(response); // Output: '2'
   }
   
   exampleWithPrompt();

4. Generate a response using messages array:

   import {generate} from '@baseai/core';
   
   async function exampleWithMessages() {
   	const response = await generate({
   		model: 'gpt-3.5-turbo-0125',
   		messages: [
   			{role: 'system', content: 'You are a helpful assistant.'},
   			{role: 'user', content: 'Give me 5 title ideas'},
   			{role: 'assistant', content: 'Sure, here you go … …'},
   		],
   	});
   
   	console.log(response);
   }
   
   exampleWithMessages();

API Reference

generate

Generates a response using the specified model, prompt, or messages array.

Signature

async function generate(params: GenerateParams): Promise<string>;

Parameters

• params: An object containing the following properties:
  • model (string): The model to use for generating the response.
  • prompt (optional string): The prompt to use for generating the response. Either prompt or messages must be provided.
  • messages (optional Message[]): An array of message objects. Each message object should contain role and content properties. Either prompt or messages must be provided.

Returns

• A promise that resolves to a string containing the generated response.

Example

import {generate} from '@baseai/core';

const responseFromPrompt = await generate({
	model: 'gpt-3.5-turbo-0125',
	prompt: '1+1',
});

console.log(responseFromPrompt);

const responseFromMessages = await generate({
	model: 'gpt-3.5-turbo-0125',
	messages: [
		{role: 'system', content: 'You are a helpful assistant.'},
		{role: 'user', content: 'Give me 5 title ideas'},
		{role: 'assistant', content: 'Sure, here you go … …'},
	],
});

console.log(responseFromMessages);

validateInput

Validates the input parameters and environment variables.

Signature

function validateInput(params: GenerateParams): ValidatedParams;

Parameters

• params: An object containing the following properties:
  • model (string): The model to use for generating the response.
  • prompt (optional string): The prompt to use for generating the response.
  • messages (optional Message[]): An array of message objects.

Returns

• An object containing the validated parameters and environment variables.

Example

const validatedParams = validateInput({
	model: 'gpt-3.5-turbo-0125',
	prompt: 'Hi',
});

buildMessages

Constructs the messages array using the provided prompt or messages array.

Signature

function buildMessages({
	prompt,
	messages,
}: {
	prompt?: string;
	messages?: Message[];
}): Message[];

Parameters

• prompt (optional string): The prompt to use for generating the response.
• messages (optional Message[]): An array of message objects.

Returns

• An array of message objects.

Example

const messages = buildMessages({prompt: 'Hi'});

buildHeaders

Constructs the headers for the API request using the provided API key.

Signature

function buildHeaders(API_KEY: string): Record<string, string>;

Parameters

• API_KEY (string): The API key to use for the request.

Returns

• An object containing the headers for the API request.

Example

const headers = buildHeaders('your-api-key');

handleResponse

Processes the API response and extracts the generated message content.

Signature

async function handleResponse(response: Response): Promise<string>;

Parameters

• response (Response): The response object from the API request.

Returns

• A promise that resolves to a string containing the generated message content.

Example

const content = await handleResponse(response);

Types

GenerateParams

Type definition for the parameters of the generate function.

interface GenerateParams {
	model: string;
	prompt?: string;
	messages?: Message[];
}

Message

Type definition for a message object.

interface Message {
	role: 'system' | 'user' | 'assistant';
	content: string;
}

ValidatedEnv

Type definition for the validated environment variables.

interface ValidatedEnv {
	API_KEY: string;
	API_URL_CHAT: string;
}

Environment Variables

OPEN_AI_API_KEY

The API key for authenticating requests to the OpenAI API.

OPEN_AI_API_URL_CHAT

The URL for the OpenAI API chat endpoint.

Example Usage

import {generate} from '@baseai/core';

const responseFromPrompt = await generate({
	model: 'gpt-3.5-turbo-0125',
	prompt: '1+1',
});

console.log(responseFromPrompt);

const responseFromMessages = await generate({
	model: 'gpt-3.5-turbo-0125',
	messages: [
		{role: 'system', content: 'You are a helpful assistant.'},
		{role: 'user', content: 'Give me 5 title ideas'},
		{role: 'assistant', content: 'Sure, here you go … …'},
	],
});

console.log(responseFromMessages);

This documentation provides a comprehensive guide for getting started with the @baseai/core package, as well as a detailed API reference for the generate function and its related components.