Skip to content
Agent Academy logo Agent Academy

Setting up your agent environment

Before you can start working with an AI coding agent, you need a properly configured environment. This module walks you through installing and configuring two agents — OpenCode (a terminal-based agent you run locally) and Codex (a cloud-based agent that runs tasks asynchronously). You will also set up the supporting tools that make agent workflows smooth: a capable terminal, clean git practices, and a sandbox project for safe experimentation.

By the end of this module, you will have at least one working agent and the environment to use it effectively.

What you will learn

Section titled “What you will learn”
  • Install and configure OpenCode for local, interactive agent workflows
  • Set up Codex for cloud-based, asynchronous task execution
  • Configure provider credentials and proxy endpoints for model access
  • Prepare your terminal environment for agent-assisted development
  • Apply git hygiene practices that work well with AI coding agents
  • Create a sandbox project for safe experimentation
  • Verify each tool is working before moving on

Prerequisites

Section titled “Prerequisites”
  • A working terminal (macOS Terminal/iTerm2, or a Linux terminal emulator)
  • Git installed and configured with your identity (git config --global user.name and user.email)
  • A code editor you are comfortable with (VS Code, Neovim, etc.)
  • An API key or account for at least one model provider (Anthropic, OpenAI, or a proxy gateway)
  • Node.js 18+ and npm (for OpenCode installation)

How agents connect to providers

Section titled “How agents connect to providers”

Both OpenCode and Codex need access to a large language model to function. The connection path differs depending on whether the agent runs locally or in the cloud, and whether your organization uses a proxy or gateway for model access.

flowchart TB
    A[You]:::neutral --> B[OpenCode]:::primary
    A --> C[Codex]:::success
    B --> D{Proxy?}:::neutral
    D -->|Direct| E[Model Provider API]:::secondary
    D -->|Via proxy| F[Proxy / Gateway]:::warning
    F --> E
    C --> G[OpenAI Cloud]:::secondary

    classDef primary fill:#dbeafe,stroke:#2563eb,color:#000
    classDef secondary fill:#f3e8ff,stroke:#7c3aed,color:#000
    classDef success fill:#dcfce7,stroke:#16a34a,color:#000
    classDef warning fill:#fef3c7,stroke:#d97706,color:#000
    classDef neutral fill:#f3f4f6,stroke:#6b7280,color:#000

Architecture diagram showing how agents connect to model providers. OpenCode connects either directly to a model provider API or through an optional proxy gateway. Codex connects to the OpenAI cloud infrastructure.

Module structure

Section titled “Module structure”

This module is organized into three sections, each with its own page:

  1. OpenCode setup — Install OpenCode, configure your model provider and proxy endpoint, set up workspace configuration, and verify the installation.
  2. Codex setup — Create your Codex account, understand the cloud execution model, connect a repository, configure autonomy levels, and run your first task.
  3. Environment essentials — Configure your terminal for agent workflows, establish git hygiene practices, and create a sandbox project for safe experimentation.

Each section includes verification steps so you can confirm your setup works before moving on.

Key takeaways

Section titled “Key takeaways”
  • A working agent environment requires three things: the agent itself, valid model provider credentials, and a project to work in
  • OpenCode runs locally in your terminal and gives you interactive, real-time control over the agent
  • Codex runs in the cloud and processes tasks asynchronously against connected repositories
  • Every setup step in this module includes a verification check — do not skip these
  • Git hygiene is more important with agents because they generate commits and modify files at a pace that demands clean history practices
  • A sandbox project gives you a safe space to experiment without risking production code

Next steps

Section titled “Next steps”