“Make this better.”
The task, audience, boundaries, and desired result are unknown. LEARNING MODE
0 OF 9 LESSONS COMPLETE
START HERE / PROMPT FUNDAMENTALS
What is a
prompt?
A prompt is the complete set of instructions and information you give an AI system to shape what it does and how it responds.
A PROMPT IS MORE THAN A QUESTION.It can define a role, task, context, boundaries, examples, quality checks, and the form of the final answer.
IntentWhat you need＋ContextWhat AI should know＋ControlsWhat must be true＋OutputHow it should arrive＝Useful prompt
“Rewrite this safety notice for new employees. Preserve every requirement. Use plain language and return a one-page checklist.”
The objective, audience, guardrail, and output are explicit.LEARN → PRACTICE → VERIFY
Build understanding,
one decision at a time.
Progress is saved in this browser. A lesson is completed only after a correct answer. Earlier self-marked records are retained separately.
Foundation · LESSON 1 OF 9
Prompt vs. question
A prompt is an instruction or question, often accompanied by context. A question such as ‘What is a prompt?’ is already a prompt. For a specific deliverable, describe the result and include information that changes the answer. You do not need a long template for every request.
Help me with this email.
Shorten the email below to 120 words for a busy client. Preserve the delivery date and requested action. [Paste email]
COACHING STUDIO / PROMPT TUTOR
Learn the pattern.
Practice with purpose.
Build prompting skill across platforms with plain-language lessons, model notes, worked examples, and a local prompt review. Guidance is teaching support, not a guarantee of output quality.
Beginner focusStart with one task and one clear deliverable. Add context only when it changes the answer.
Review a promptFOUNDATIONS · BEGINNER
Outcome-first instructions
Name the action and the result a person should be able to use. Replace vague requests such as “help me” with a concrete verb and deliverable.
WORKED PATTERN
Summarize the attached meeting notes into decisions, open questions, and next actions.
Try the example once, then note what information was missing or unnecessary.
MODEL-SPECIFIC ADAPTATION
Keep the goal. Adapt to the system.
Prompt patterns transfer, but product interfaces, message roles, tools, output controls, and model behavior can differ. Check the source documentation for the exact model and version.
GUIDANCE
Start with the task, context, boundaries, output, and a check. Keep this stable when you compare systems.
TRY THIS STRUCTURETask: [specific result] Context and sources: [what matters] Boundaries: [what to preserve, avoid, or escalate] Output: [usable format] Check: [observable acceptance criteria]Shared prompting baseline
OpenAI ChatGPT and GPT modelsOfficial provider referenceAnthropic ClaudeOfficial provider referencexAI GrokOfficial provider referenceGoogle GeminiOfficial provider referenceMicrosoft CopilotOfficial provider referenceMeta Llama and open-weight modelsOfficial provider referenceMistralOfficial provider referenceDeepSeekOfficial provider reference
COACHED PRACTICE / LOCAL CHECK
Review and improve a prompt.
Your draft stays in this page and is not sent to an AI service. The checklist looks for signals in the text; it cannot judge truth, fit, safety, or model performance.
REVIEW CHECKLIST
- +A concrete task or outcomeConsider adding
- +Useful audience or contextConsider adding
- +A boundary or source ruleConsider adding
- +A usable output shapeConsider adding
- +A way to verify the resultConsider adding
These are text-pattern checks, not an automated quality score. A missing keyword may be present in different wording; a found keyword does not prove the requirement is met.
COACHING SUGGESTION
Enter a prompt to receive a transparent checklist and a copyable working version with placeholders for details that may need clarification.
What this tutor can and cannot establish
It teaches general techniques and links to provider documentation. Its local checklist is intentionally heuristic. It does not connect to ChatGPT, Claude, Grok, Gemini, Copilot, Meta, Mistral, DeepSeek, or another model; it does not verify current model availability, execute prompts, or certify that an output is correct. For production workflows, validate with your actual product surface, model version, permissions, representative data, and human review.
PRACTICE / PROMPT BUILDER
Build a prompt.
Watch it take shape.
Use the fields that matter to your task. Role is optional. Filling every field does not prove quality. Start from scratch or adapt a practical template.
STARTING POINT
PRACTICE / SCENARIO CHALLENGES
Make the decision.
See the system logic.
Apply prompt architecture to realistic training, analysis, operations, and governance decisions.
The unsupported safety module
You have meeting notes and an outdated checklist. A manager asks AI to create final safety training by tomorrow.
KNOWLEDGE CHECK / 01
Where does this instruction belong?
“Do not invent numbers. Flag missing evidence for human review.”