Getting Started

Get up and running with Coding Sandbox in under five minutes.

Prerequisites

Before you begin, make sure you have the following installed on your machine:

Docker Desktop (or Docker Engine on Linux) — Coding Sandbox uses Docker to create isolated environments.
An internet connection — required for downloading agent templates and activating tunnels.

Docker required

Coding Sandbox relies on Docker to create and manage containers. Make sure Docker Desktop is running before launching the app.

Create Your First Sandbox

Once Coding Sandbox is installed and running, creating a sandbox takes just one click.

Select an Agent

Click the + New Sandbox button in the dashboard. You'll see a list of supported AI coding agents:

Claude (Anthropic)
Codex (OpenAI)
Copilot (GitHub)
Gemini (Google)
Kiro (AWS)
OpenCode (open-source)
CAgent (Docker)

Select the agent you want to work with. Each agent comes with a pre-configured Docker environment optimized for that specific tool.

Launch the Sandbox

After selecting an agent, click Create. Coding Sandbox will:

Pull the appropriate Docker image
Configure the environment with sensible defaults
Start the container
Optionally set up a secure Cloudflare tunnel

You can also create sandboxes from the CLI:

bash

csb create --agent claude --name my-project

Connect Your Agent

Once the sandbox is running, you can connect your AI coding agent to it. The sandbox exposes the necessary ports and provides a secure URL if tunnel access is enabled.

bash

# View your running sandboxes
csb list

# Get the connection URL for a sandbox
csb info my-project

Pro tip

Enable tunnel access to get a public URL for your sandbox — perfect for connecting cloud-hosted AI agents.

Next Steps

Learn about Sandbox Management for advanced container operations
Set up Tunnel Access for secure remote connections
Explore the CLI Reference for the full list of commands

Introduction

Installation