Skip to content
ai-gent

Quick Start

This guide walks you through creating a simple agent that answers questions using a local LLM.

Prerequisites

Section titled “Prerequisites”
  • Ollama installed and running
  • A model pulled (e.g., ollama pull qwen3:8b)
  • Node.js 20+ or Python 3.11+

Installation

Section titled “Installation”
Terminal window
# Coming soon — the framework is under active development
npm install @ai-gent/core

Your First Agent

Section titled “Your First Agent”
import { Agent, OllamaProvider } from '@ai-gent/core';


const agent = new Agent({
  provider: new OllamaProvider({
    model: 'qwen3:8b',
    baseUrl: 'http://localhost:11434',
  }),
  tools: [],
  systemPrompt: 'You are a helpful assistant.',
});


const result = await agent.run('What is the capital of France?');
console.log(result);

Adding Tools

Section titled “Adding Tools”

Tools give your agent superpowers. Here’s a simple web search tool:

import { Agent, OllamaProvider, defineTool } from '@ai-gent/core';


const searchTool = defineTool({
  name: 'web_search',
  description: 'Search the web for information',
  parameters: {
    query: { type: 'string', description: 'Search query' },
  },
  execute: async ({ query }) => {
    // Your search implementation here
    return `Results for: ${query}`;
  },
});


const agent = new Agent({
  provider: new OllamaProvider({ model: 'qwen3:8b' }),
  tools: [searchTool],
});

Using LiteLLM Router

Section titled “Using LiteLLM Router”

For hybrid local/cloud routing:

import { Agent, LiteLLMProvider } from '@ai-gent/core';


const agent = new Agent({
  provider: new LiteLLMProvider({
    baseUrl: 'http://localhost:4000/v1',
    apiKey: 'your-litellm-key',
    model: 'local-general', // Routes via LiteLLM
  }),
});

Next Steps

Section titled “Next Steps”