Quick Start

In this guide, you will install the SDK, configure your first Language Model provider, and build a simple agent capable of calling a tool to fetch weather data.

Prerequisites

• Node.js v18 or later
• TypeScript configured in your project
• An API key for OpenAI, Groq, Gemini, or Anthropic

Step 1: Install the SDK

Install the core umbrella package via your preferred package manager. It includes everything you need: the agent runtime, memory, guardrails, and all officially supported providers.

npm install @rabbit-agent-sdk/rabbit-agent-sdk zod

(We also install zod as it is required for defining tool schemas).

Step 2: Initialize the Agent

Create a new file called agent.ts. We'll initialize an agent using the OpenAIProvider.

import { Agent, OpenAIProvider } from "@rabbit-agent-sdk/rabbit-agent-sdk";
 
// The agent requires a provider to communicate with the LLM.
const agent = new Agent({
  provider: new OpenAIProvider({ 
    apiKey: process.env.OPENAI_API_KEY // Ensure this env var is set
  }),
  systemPrompt: "You are a helpful and concise assistant.",
});
 
// Run the agent!
const response = await agent.run("Hello, who are you?");
console.log(response);

Run this script using ts-node or tsx (e.g. npx tsx agent.ts) and you'll get a greeting from the LLM.

Step 3: Add a Tool

Agents become powerful when they can interact with the outside world. Let's give our agent a getWeather tool using Zod for type-safe arguments.

import { Agent, OpenAIProvider, Tool } from "@rabbit-agent-sdk/rabbit-agent-sdk";
import { z } from "zod";
 
const weatherTool: Tool = {
  name: "getWeather",
  description: "Get the current weather for a specific location.",
  schema: z.object({
    location: z.string().describe("The city and state, e.g., San Francisco, CA"),
  }),
  execute: async ({ location }) => {
    // In a real app, you would call a weather API here.
    return `The weather in ${location} is sunny and 75°F.`;
  },
};
 
const agent = new Agent({
  provider: new OpenAIProvider({ apiKey: process.env.OPENAI_API_KEY }),
  systemPrompt: "You are a helpful assistant. Use tools when necessary.",
  tools: [weatherTool], // Register the tool
});
 
const response = await agent.run("What's the weather in Tokyo?");
console.log(response);
// → "The weather in Tokyo is sunny and 75°F."

Behind the scenes, the SDK:

1. Converted your Zod schema into a JSON Schema.
2. Sent it to OpenAI.
3. Received the tool call request from the model.
4. Validated the arguments.
5. Executed your execute function and fed the result back to the LLM to formulate the final answer.

Next Steps

You now have a working agent! Dive deeper into the core concepts to see what else Rabbit SDK can do:

• 🔌 Providers: Learn how to switch to Groq, Gemini, or Claude.
• 🛠️ Tools: Build more complex tools and use the built-in toolkit.
• 🧠 Memory: Manage conversational history.
• 🛡️ Guardrails: Protect your agent against prompt injection and enforce output schemas.