The New Skills Stack: Requirements → Specs → Structured Prompts → Agents

STAGE 1: REQUIREMENTS. What does the software need to do? What problem does it solve? Who uses it? What does success look like? This is a business and product skill, not a technical one. You can write requirements in plain English with no technical knowledge. The skill is clarity, not jargon.

STAGE 2: SPECIFICATIONS. How specifically does the software need to do it? What are the inputs and outputs? What are the rules? What are the edge cases? What happens when something goes wrong? This is where business language becomes engineering language. The skill is precision and completeness.

STAGE 3: STRUCTURED PROMPTS. The specification, formatted in a way the agent can ingest cleanly. Markdown headers, code fences for examples, bullet lists for rules, explicit "here is what's in scope" and "here is what's out of scope" sections. The skill is structure and discipline.

STAGE 4: AGENT EXECUTION. The agent reads the prompt, writes code, and hands it back. You review, run, and iterate. The skill is verification — knowing when the output is acceptable and when it needs another pass.

Most beginners try to do all four stages at once, in their head, while typing into a chat box. That's why their agents produce garbage. They skipped from "I have an idea" straight to "type something at the agent."

The discipline this curriculum teaches: separate the four stages. Write requirements first. Convert them to specs. Format the specs into structured prompts. Then, and only then, talk to the agent.

We'll cover each stage in detail in Part 2.