OpenCode Masterclass: AI Agents, MCP, Skills & Subagents

Stop paying for expensive AI coding subscriptions. Start building your own AI-powered coding workflow.

OpenCode is an open-source, provider-agnostic AI coding agent that runs directly in your terminal. This practical crash course takes you from your first OpenCode command to building a structured, reusable agentic coding workflow using Slash Commands, Context Management, AGENTS md, MCP, Skills, and Subagents.

This is not a course about simply typing prompts and hoping for the best. You’ll learn how to turn OpenCode into a reliable AI development assistant that understands your project, uses external tools, follows your instructions, and delegates specialized work to agents

What You’ll Build

Throughout the course, you’ll work with practical, hands-on examples and progressively build your own OpenCode toolkit:

• Custom Slash Commands to automate repetitive tasks and stop rewriting the same prompts

• A project-aware AGENTS md file that gives OpenCode persistent instructions about your project

• MCP connections to GitHub, databases, and documentation using real hands-on examples

• A Data Cleaning Skill for handling real-world data problems

• A Data Migration Skill for specialized data workflows

• A Dashboard Generation Skill for turning data into visual reports

• A custom Code Reviewer Subagent that can automatically review your work

• A custom Orchestrator Subagent for coordinating specialized tasks

What You’ll Learn

• What OpenCode is and how agentic coding differs from traditional coding

• How the OpenCode agentic loop works internally

• How to set up OpenCode using free models, plus optional paid configuration

• How to create and use Slash Commands for repeatable workflows

• How Sessions and Resume work when working across longer tasks

• How OpenCode's Context Window works and why context management matters

• Auto-compaction and practical techniques for keeping long AI sessions focused

• How to write, test, and maintain an AGENTS md file

• What MCP (Model Context Protocol) actually is and how it connects OpenCode to external tools

• Hands-on MCP integrations with GitHub, databases, and Context7

• MCP Tools, Permissions, and the Allow / Ask / Deny trust model

• The most useful MCP servers developers should know, including Notion, Slack, Jira, AWS, Docker, and Filesystem

• How OpenCode Skills work and why they are useful for specialized workflows

• Skills architecture, including YAML front matter and Markdown instructions

• How to create your own Data Cleaning, Data Migration, and Dashboard Generation Skills

• How to install Skills from the OpenCode Marketplace

• What Subagents are and how OpenCode's built-in agents work

• How to use the built-in Explore, Plan, and General-Purpose subagents

• How to build your own Custom Subagents for specialized workflows

• Skills vs Subagents — when to use each and how they complement one another

• How to build an Orchestrator Agent that coordinates complex workflows

Who This Course Is For

This course is designed for developers, data analysts, data scientists, engineers, freelancers, and AI enthusiasts who want to move beyond basic AI prompting and learn how to build practical agentic coding workflows.

It’s especially useful if you want to:

• Automate repetitive coding and data tasks

• Build reusable AI-powered workflows

• Connect your coding agent to external tools and data

• Create specialized Skills and Subagents

• Understand context management and improve AI reliability

• Learn how modern agentic coding systems actually work

You don’t need to be an AI expert. If you’re comfortable working with a terminal and have basic programming knowledge, you can follow along and start building your own OpenCode workflows.