The vibecoding workflow
Good vibecoding isn't about finding the perfect prompt — it's about managing a repeatable workflow: give the agent the right context, agree on what to build, verify the result, and preserve what you learn for next time.
In Module 1, we focused on how to communicate with an LLM. Here, we move one level up: how to manage the work itself.
For a small task, you may be able to describe what you want and start immediately. Larger projects need more structure. The agent has to understand the project, the task may need to be refined before building, and the result needs to be verified before you rely on it.
What you'll be able to do
Take a project from an initial idea to a verified result using a repeatable workflow. You'll learn when to plan before building, how to organize the information an agent needs, how to specify tasks without micromanaging the implementation, how to verify the result, and how to preserve useful context across sessions.
How this module is structured
Eight sub-pages, in order:
1. Plan before you build
Why letting the agent investigate and propose a plan before it touches anything saves more time than it costs.
2. Set up your information environment
A clear project folder, telling the agent what's in it, and a project guide for larger projects.
3. Specify the task
How specific to be, and how to break a complicated task into steps with the right human checkpoints.
4. Verify before you trust
A four-question framework for checking that output isn't just polished, but actually correct.
5. Carry context across sessions
Workflows, rules, corrections, and progress — and how project instruction files and built-in memory fit in.
6. Helpful tips
Habits that make Claude Code sessions go better: protect the context window, show rather than describe, and know when to interrupt or start fresh.
7. Use case I: event recap generator
Try the task yourself, then compare three build processes — browser prompt, Plan mode, and an explicit CLAUDE.md — against the same reference recap.
8. Use case II: model automation
End-to-end: automate the monthly three-statement update without giving up the model your CFO trusts.