Getting Started

Welcome to ZERG — the production-grade agentic orchestration platform. This guide will help you go from zero to running your first AI agent swarm.

Prerequisites

• Erlang/OTP 27+ (28 recommended)
• LuaJIT 2.1+
• rebar3
• Python 3.11+ (for Mango auth service)
• PostgreSQL 16+ (for memory/embeddings)

Quick Start

1. Build the Server

cd server
rebar3 compile
rebar3 release

2. Start the Server

rebar3 shell
# or use the release
_build/default/rel/sol/bin/sol start

3. Build the CLI

cd zerg-cli
bash build/build.sh

4. Your First Agent

# Start a Luna agent (worker)
cd client
luajit main.lua --demo

# Or use the CLI
zerg status
zerg models
zerg inference start --model glm-5.1

Architecture Overview

ZERG consists of three main components:

Component	Technology	Purpose
Sol	Erlang/OTP	Orchestration server — 110+ REST endpoints, inference, workflows
Luna	LuaJIT	Worker agent — <10MB RAM, 45+ tools, 6 agent modes
Mango	Python/Tornado	Auth service — token validation, RBAC

Next Steps

• Installation Guide — detailed platform setup
• Architecture — deep dive into system design
• API Reference — HTTP API documentation
• CLI Reference — ZERG CLI commands