Prompt Engineering & Structured Output
Turn a naive prompt into a structured-output pipeline.
7 lessons 3 phases intermediate
Get started
Signed in as your account.
why this workshop
who it's for.
Developers who finished W1 and want to turn a naive prompt into typed, validated, critic-filtered output.
exam coverage
Maps to Domain 4 (Prompt Engineering & Structured Output), 20% of the CCA – Foundations exam.
what you'll build
what you walk away with.
- 01 A baseline eval that measures how bad the W1 prompt actually is.
- 02 An explicit rubric prompt with four severity levels and ignore rules.
- 03 Few-shot examples that cut the false-positive rate.
- 04 A record_findings tool schema so Claude returns typed JSON findings.
- 05 A Zod-validated schema with confidence scores and evidence ranges.
- 06 A validation-retry loop that corrects invalid output automatically.
- 07 A two-pass critic that filters low-confidence findings.
curriculum
7 lessons across 3 phases.
phase A · Diagnosis & criteria 3 lessons
- 01 Baseline
- 02 Explicit Criteria
- 03 Few-Shot Examples
phase B · Structured output 2 lessons
- 04 tool_use Structured Output
- 05 JSON Schema Design
phase C · Hardening 2 lessons
- 06 Validation-Retry Loop
- 07 Multi-Pass Review
prerequisites
what you need before you start.
- Completed W1 (or equivalent)
- The workshop reads src/review.ts as its starting state. You need it on disk.
- Node.js 22+
- Same stack as W1 — pnpm workspaces, tsx, TypeScript.
start
how to start.
Once Claude Code is set up, tell it “start the prompting workshop” and it clones the project and queues the first lesson.
New here? The
getting-started guide
walks you through installing the lwc CLI and the Claude Code
plugin once — everything you need before that line works.
the series