Agent Academy

What is Agent Academy?

Agent Academy is a free, structured curriculum that teaches developers how to work effectively with AI coding agents like OpenCode and Codex. Instead of scattered blog posts and tool docs, you get a clear learning path across nine modules — from understanding what agents are, through prompt engineering, context engineering, and agent skills, all the way to MCP servers, subagent delegation, and security hardening.

Structured path

Nine progressive modules that build on each other, so you learn in the right order.

Hands-on exercises

Every module includes practical exercises you complete in your own projects.

Tool-agnostic concepts

Core ideas apply broadly, with specific guidance for OpenCode and Codex.

Open source

All content is open source. Read it online, suggest improvements, or contribute.

Who is this for?

New to AI coding agents? Start at Module 1 to understand what agents are, how they work, and when to use them. By Module 2, you will have a working agent setup ready for real tasks.

Already using agents casually? Jump to the module that fills your gaps. Most developers benefit from Modules 3-5 on prompt engineering, context engineering, and agent skills — the techniques that turn inconsistent results into reliable workflows.

Modules

Module 1: Introduction to AI coding agents What AI coding agents are, how they work, and how they differ from chat assistants and autocomplete tools.

Module 2: Setting up your agent environment Get OpenCode and Codex installed, configured, and verified so you can start working with agents.

Module 3: Prompt engineering for coding agents Communicate effectively with agents through structured prompts that produce predictable, high-quality results.

Module 4: Context engineering Provide project-level context so your agent produces code that matches your standards without repeated instructions.

Module 5: Agent skills Use and create reusable skill modules that extend your agent's capabilities and reduce repetitive instruction.

Module 6: MCP servers Connect your agent to external tools, services, and data sources via the Model Context Protocol.

Module 7: Subagents and task delegation Delegate complex work to subagents and learn when delegation helps versus when it adds unnecessary complexity.

Module 8: Spec-Driven Development (SDD) Use a structured specification-to-code pipeline to build features systematically instead of relying on ad-hoc prompting.

Module 9: Security, guardrails, and safe automation Identify and mitigate the security risks of AI coding agents to protect your codebase and credentials.

Where to start

Recommended starting point

Module 1: Introduction is the best place to begin. It takes about 30 minutes and gives you the mental model you need for everything that follows. From there, Module 2 walks you through setting up your agent environment step by step.

If you already have an agent installed and want to improve your results, skip ahead to Module 3: Prompt engineering — it covers the techniques that make the biggest difference in day-to-day agent use.