Basic Agents
You know how to use LLMs in the browser. Now learn what agent harnesses exist, how they cost, and how to get one working for you.
What This Track Covers
6 lessons to go from competent chat user to someone who can deploy, use, and budget for autonomous agents.
Lesson 1
What Is an Agent Harness?
The difference between a chatbot and an agent. What a harness does (model routing, tool execution, memory, sandboxing). A simple mental model: the agent is the brain, the harness is the body.
Lesson 2
The Options: Desktop vs TUI Agents
Desktop-first tour of agent harnesses: Opencode Desktop and Claude Cowork. Then the TUI options. Highlights what each is built for, who it's for, and the key differences between a visual agent workspace and a terminal-driven one.
Lesson 3
How Much Does an Agent Cost Per Hour?
The real breakdown. Token pricing vs subscription pricing. What an average "hour of work" costs across different models and endpoints. Includes worked examples and a spreadsheet template.
Lesson 4
Setting Up Your First Agent
Step-by-step: pick a harness, get an API key, configure your model, give it a tool (search or file access), and watch it complete a task. Should take under 30 minutes.
Lesson 5
Paying for It: API Keys, Credits, and Subscriptions
How API billing works. Pre-paid credits vs post-paid. Rate limits, concurrency, and the hidden costs (context caching, image inputs, tool call tokens). How to set a monthly budget and not exceed it.
Lesson 6
When to Stay in the Browser
Agents aren't always the answer. A framework for deciding: can your task be done with a good prompt in the chat window? When is the overhead of an agent harness worth it, and when is it just complexity for its own sake.
Lessons being published throughout 2026.