Getting Started

Welcome to Orbiter. This section walks you from zero to a working multi-agent system.

Learning Path

Follow these pages in order for the best experience:

1. Installation

Set up Orbiter in your project. Covers installation from GitHub, UV workspace development, Python version requirements, and environment variables for LLM providers.

2. Quickstart

Build and run your first agent in under 5 minutes. A weather-bot example that covers @tool, Agent, run.sync(), streaming, and multi-turn conversations.

3. Core Concepts

Understand the building blocks of Orbiter: Agent, Tool, Runner, Swarm, message types, RunResult, and streaming events. This is the reference you will come back to most often.

4. Your First Agent

A step-by-step tutorial that builds a real multi-agent system from scratch. Define tools, create agents, inspect results, add handoffs, set up a Swarm workflow, and use structured output.

What You Will Build

By the end of this section you will know how to:

• Install Orbiter and configure LLM providers
• Define typed tools with the @tool decorator
• Create agents with instructions and tools
• Run agents synchronously, asynchronously, and with streaming
• Build multi-turn conversations by passing message history
• Orchestrate multiple agents with Swarm (workflow, handoff, and team modes)
• Validate agent output with Pydantic structured output

Prerequisites

• Python 3.11 or later
• An API key for at least one LLM provider (OpenAI or Anthropic)
• Basic familiarity with async/await in Python (helpful but not required — run.sync() provides a blocking API)

Next Steps

Once you have finished the Getting Started section, explore:

• Guides — Deep dives into context engine, memory, tracing
• Architecture — How Orbiter is designed internally
• API Reference — Complete API documentation