From prompt to agent
What changes when the AI can read your files, run code, and act in loops — and why that changes what's possible.
What you'll be able to do
Explain what an "agent" actually is, recognize when a task needs one versus a regular chatbot, and describe what vibecoding is and how it shifts your job from writing code to directing an agent — before installing anything.
How this module is structured
Three sub-pages, in order:
1. The agent mental model
How chat evolved into agents through reasoning, tool use, and the ReAct loop — and a side-by-side comparison of what changes.
2. What agents unlock
An introduction to vibecoding: what it is, how the job shifts from implementing to directing, and principles for doing it well.
3. The agent harness
What surrounds the model to make it useful for a job: instructions, context, tools, and boundaries.