Vague or generic output
Add a clear task, audience, context and output format. Start with Elements.
PROMPT ENGINEERING, FROM TASK TO EVALUATION
Prompt Engineering Tutorial and Prompt Practice
Stop collecting magic phrases. Learn to define the task, supply useful context, constrain the output and test whether the prompt still works on the next example.
102 chapters8 core gatesPrompt Playbook
prompt.playbookEVAL READY
TASKTurn five interview notes into a product decision.
CONTEXTAudience, source notes, definitions and known gaps.
CONSTRAINTCite each claim. Mark missing evidence. Do not infer quotes.
OUTPUTDecision table: evidence, confidence, risk, next test.
RUN 01
EDGE CASE
HUMAN REVIEW
Why this page does not show 102 chapters at once
JR Academy has 102 Prompt Engineering chapters. A flat directory makes the first decision harder, so this page identifies eight core gates and keeps the remaining lessons in a collapsible reference library. Every existing chapter URL remains unchanged.
Choose the right Prompt technique for the failure you actually have
Most users do not need an advanced framework first. Diagnose the output, then add only the missing structure.
Format changes every run
Add examples and a validation rule. Move to Few-shot and Evaluation.
The task needs several decisions
Split it into a chain and keep each handoff inspectable.
Eight gates produce one Prompt Playbook
Use one real task throughout the route. Each gate adds a reusable prompt, test result or safety rule to the same playbook.
01 · Choose
Real task and success condition
Complete and save this evidence in the chapter workspace.
02 · StructureFour-part Prompt Brief
Complete and save this evidence in the chapter workspace.
03 · BaselineBaseline output and gap log
Complete and save this evidence in the chapter workspace.
04 · TeachFew-shot example set
Complete and save this evidence in the chapter workspace.
05 · ChainPrompt workflow map
Complete and save this evidence in the chapter workspace.
06 · EvaluateTest set and scorecard
Complete and save this evidence in the chapter workspace.
07 · ProtectInjection boundary and fallback
Complete and save this evidence in the chapter workspace.
08 · StandardiseReusable system prompt contract
Complete and save this evidence in the chapter workspace.
Use the library when a specific problem appears
The core route teaches the method. These entry points help you solve writing, extraction, coding, reasoning, image and Agent problems without browsing 102 titles.
REFERENCE ENTRY
Writing and summary
Control audience, length, evidence and output shape.
REFERENCE ENTRYExtraction and structured output
Turn documents into fields that can be checked or processed.
REFERENCE ENTRYCoding prompts
Specify environment, constraints, tests and acceptable changes.
REFERENCE ENTRYReasoning tasks
Choose decomposition and verification without requesting hidden reasoning.
REFERENCE ENTRYImage generation
Control subject, composition, light, text and destination.
REFERENCE ENTRYWorkflows and agents
Know when a prompt is enough and when the system needs state or tools.
Complete Prompt Engineering chapter directory
All existing chapters remain linked for search discovery and returning learners. Groups stay collapsed until you need them.
Introduction6 chapters
Techniques17 chapters
Prompt Library36 chapters
24 · Prompt Library25 · Classification Prompts26 · Sentiment Classification27 · Sentiment (Few-shot)28 · Adversarial Prompts29 · Prompt Leaking30 · Jailbreaking31 · Mathematics Prompts32 · Adding Odd Numbers33 · Composite Functions34 · Reasoning Prompts35 · Indirect Reasoning36 · Physical Reasoning37 · Coding Prompts38 · Code Snippets39 · MySQL Query40 · TikZ Diagram41 · Q&A Prompts42 · Open Domain Q&A43 · Closed Domain Q&A44 · Science Q&A45 · Information Extraction46 · Extract Entities47 · Text Summarization48 · Explain a Concept49 · Image Generation50 · Alphabet Person (TikZ)51 · LLM Evaluation52 · Evaluate Outputs (Teacher)53 · Truthfulness54 · Identify Hallucination55 · Creativity56 · Interdisciplinary Ideas57 · Infinite Primes (Shakespeare)58 · Invent New Words59 · Rhyming Proofs
Agents6 chapters
Models20 chapters
System Prompts Masterclass8 chapters
Google Official Essentials5 chapters
Prompt Playbook
Leave with prompts your team can review and reuse
A Prompt Playbook records more than the final wording. It preserves the task, inputs, test cases, failure modes and handoff rules.
Task and audience brief
Reusable prompt template
Few-shot example set
Prompt chain map
Evaluation scorecard
Failure and safety rules
Version decision log
System prompt contract
Prompt Engineering questions people ask before starting
Is Prompt Engineering still worth learning?
Yes, but the useful skill is no longer memorising keywords. Simple questions need less technique; structured output, batch processing, tool use and team workflows still need someone to design the task boundary, input sources, output contract and evaluation cases. This track teaches those transferable decisions.
Do I need to know programming to learn Prompt?
No. Prompting is essentially describing tasks in natural language - the key is breaking down requirements clearly. However, knowing some JSON and API concepts helps when you reach the Agent and Tool Use sections.
Will these Prompt methods become obsolete when models change?
Model-specific parameters and preferences change. The engineering decisions around task, context, constraints, output format, examples and evaluation do not disappear together. The track separates transferable methods from model-specific tactics; when you switch models, rerun the test set before tuning details.
What is Prompt Lab?
Prompt Lab is an interactive practice area. Write a prompt for a defined task, inspect the result, then revise it against output format, factual and constraint checks. It is designed for method validation and does not require coding first.
How long does it take to complete?
The core route has eight gates. Practise one real task for 30 to 60 minutes a day, complete the foundation first, then open the writing, coding, image or Agent library when the work requires it. Completion means delivering a tested Prompt Playbook, not reading all 102 chapters.
Content reviewed by Lightman WangFounder of JR Academy · JR Academy
LinkedIn ↗