Documentation
Docs
Getting Started

Quickstart Guide

Go from zero to a production-grade multi-agent system in less than 5 minutes.

Prerequisites

  • Docker Desktop or automatic Kubernetes cluster (Kind/Minikube)
  • Helm 3+
  • kubectl

1. Install the CLI

The Consonant CLI (`cons`) is your main interface for managing agents and workflows.

bash
# Install via Homebrew
brew install consonant-ai/tap/consonant

# Verify installation
cons version

2. Initialize the Control Plane

This command spins up the Consonant control plane in your current Kubernetes context. It installs the Manager, the Ingress Controller, and the Observability stack.

bash
cons init --local
Local Mode
The --local flag optimizes resource usage for laptop development. For production clusters, omit this flag.

3. Create a Project

Generate a new project with sample agents.

bash
cons create my-first-project
cd my-first-project

This creates a directory structure with:

  • consonant.yaml: The project manifest
  • agents/: Directory for your agent code
  • policies/: OPA policies

4. Deploy Agents

Deploy the agents defined in your project to the cluster.

bash
cons deploy

You should see pods spinning up:

bash
kubectl get pods -n consonant-agents

5. Run Your First Goal

Submit a goal to the control plane. The planner will automatically route it to the appropriate agent.

bash
cons run --goal "Research the history of the diode" --stream

You'll see a real-time stream of the execution plan and the agent's output.

Next Steps

Now that you have a running system, try these next steps:

Core Concepts

Understand Projects, Agents, and Goals.

Create Custom Agents

Learn how to package your own code as a KAgent.