Module 02 · Base

Prompt engineering

The parts of a working prompt, what examples teach that instructions cannot, how to get output a program can branch on, and how to measure a prompt change instead of guessing at it.

4 lessons · 3 graded exercises·3.5 hours·beginner·Prerequisite: LLM fundamentals

Anatomy of a prompt

25 min

A working prompt has five parts, each doing a different job to the distribution. Where they go, how specific to be, and which of the famous tricks are load bearing.

Teaching with examples

30 min · exercise

An example carries format and edge case handling that no instruction states cleanly. How many to use, which ones to pick, and what an accidental shared property teaches by mistake.

Structured output

35 min · exercise

Three ways to get machine readable output and what each costs. Common parse failures, validation as the thing that makes failure visible, and the retry that feeds the error back.

Measuring a prompt

40 min · exercise

A prompt change you did not measure is a guess. Ten honest cases, graders that are deterministic, pass rate as a comparison between two versions, and regression as the alert worth having.