Skip to main content
Welcome to the localtrain.ai Typescript SDK! A powerful TypeScript toolkit designed to help you seamlessly build AI-powered agents using any LLMs on runtimes like Node.js. Explore the SDK to integrate AI into your applications effortlessly and scale your capabilities!
Before you dive in, make sure you have Node.js installed on your machine.

​
Installation

First, install the Core Package of localtrain.ai:

​
Using npm

npm install @localtrain-ai/core

​
Using yarn

yarn add @localtrain-ai/core
Now, you’re ready to start building amazing AI-powered solutions! 💡

​
📚 Usage Guide

​
Example: Building an AI Agent with the SDK Core

Let’s walk through creating a Customer Feedback Analyzer Agent step by step:

​
Step 1: Initialize the SDK

Start by importing the required components and initializing the localtrain.ai SDK: 
import { LocalTrain, Agent, ContextManager, IAgent } from '@localtrain-ai/core';
import { Openai } from '@localtrain-ai/llm/openai';

async function main() {
    // Initialize the SDK
    LocalTrain.initialize();
}
👉 What’s happening here? The LocalTrain.initialize() sets up the SDK, making it ready to use.

​
Step 2: Attach Providers to the Registry

Add the necessary providers (e.g., OpenAI) for your agent to use: 
import { LocalTrain, Agent, ContextManager, IAgent } from '@localtrain-ai/core';
import { Openai } from '@localtrain-ai/llm/openai';

async function main() {
    // Initialize the SDK
    LocalTrain.initialize();

    // Attach providers to the provider registry
    LocalTrain.useProviders(
        // Add OpenAI as a provider
        new Openai({
            apiKey: process.env.OPENAI_API_KEY // Pass your OpenAI API Key
        }),
        // You can add more providers here, separated by commas
    );
}
👉 What’s happening here? The LocalTrain.useProviders() function registers the providers, allowing your agent to leverage their services (e.g., OpenAI for LLM functionalities).

​
Step 3: Create and Configure the Agent

Define an agent with its configuration and purpose: 
    // Create a new agent instance
const feedbackAgent = new Agent({
    contextManager: new ContextManager().setValue("appId", "feedback-analyzer"),
});

// Define the agent details
const agentDetails: IAgent = {
    name: "Customer Feedback Analyzer",
    description: "Analyzes customer feedback and provides insights.",
    appId: "feedback-agent-id",
    inputs: [
        {
            name: 'feedback',
            title: 'Customer Feedback Input',
            type: 'Text',
            options: [],
        }
    ],
    behaviours: [
        {
            behaviourId: 'sentiment-analysis',
            name: 'AnalyzeFeedback',
            provider: 'open-ai',
            inputs: {
                prompt: `You are a customer feedback analyzer. Your role is to process customer feedback and provide insights such as:
                            - Key themes (e.g., product issues, feature requests)
                            - Sentiment (positive, negative, neutral)
                            - Actionable recommendations
                            Feedback: @feedback`,
                temperature: 0.7, // Adds variability for creative summarization
                model: "gpt-4o-mini", // Specify the model
                responseType: "text",
            },
            behaviours: [],
        }
    ]
} as IAgent;

// Add the configuration to the agent
feedbackAgent.addConfig(agentDetails);
👉 What’s happening here? You’re creating an agent feedbackAgent and specifying its name, description, and behaviors. Each behavior defines how the agent interacts with the OpenAI provider.

​
Step 4: Run the Agent

Finally, run the agent with a custom prompt: 
const feedback = "The app is great, but I wish the loading time was faster. Also, adding dark mode would be helpful.";
const result = feedbackAgent.run({ feedback });

console.log("Feedback Analysis Result:", result);

main();
👉 What’s happening here? The run() method executes the agent with the given prompt and logs the results.

​
Ready to build your first AI-powered agent? 🚀 Dive into the Documentation and start exploring today!

​
🌐 Community

Join our community to connect, share ideas, and get support: We’d love to hear from you and help you build amazing projects!

​
📜 License

This project is licensed under the MIT License. See the LICENSE file for details.