free field guide workflows

15 ChatGPT Codes for Studying

Use fifteen custom prompt labels for source-grounded study, recall, diagrams, practice questions, reasoning, and plain-language explanations.

15 ChatGPT Codes for Studying

These slash labels are custom prompt shortcuts, not built-in ChatGPT commands. The full instruction behind each label is what makes it work.

What this helps you do

Use one approved source to find missing foundations, test what you remember, compare confusing concepts, build a diagram, and write realistic practice questions.

Before you start

Gather your learning objectives, instructor notes, slides, or approved textbook excerpt. Remove private student or patient information. Decide whether outside facts are allowed. For graded work, follow your school's AI policy.

Terms to know

Source-grounded: every answer must trace back to material you supplied.
Prerequisite: knowledge you need before the current topic will make sense.
Contrast table: a side-by-side comparison built around meaningful differences.
Retrieval practice: trying to recall information before looking at the answer.

Step-by-step

Start a new chat and upload or paste the approved course material.
Paste the setup block below.
Run one code at a time.
Answer questions before asking ChatGPT for feedback.
Check every explanation against the original material.

Copy this

Use only the course material I provide. Cite the source filename and heading for every answer. If the material does not support an answer, write `Not in the source`.

/PREREQUISITE: Before teaching this topic, identify the five ideas I must already understand. Ask me one short diagnostic question for each. Wait for my answers, then tell me what to review first.

/BRAINDUMP: Ask me to explain the topic from memory without notes. After I answer, compare my response with the source. List what I got right, what I missed, and what I stated incorrectly.

/CONTRAST: Put the two concepts I name in a table. Compare definition, purpose, mechanism, key signs, common confusion, and one example. Use only supported rows.

/VISUALIZELEARNING: Turn the process into a simple labeled diagram with arrows in the correct order. Under the diagram, explain each step in one sentence and cite its source.

/QUESTIONPREDICTOR: Using only the learning objectives and source, write five likely quiz questions. Ask them one at a time. Wait for my answer, then show the correct answer, explanation, and source line.

Treat these as custom instructions, not native commands. Follow a code only when I type it.

Five reasoning codes from the newer reel

Copy this

/STEELMAN: Argue the strongest version of the opposite position, then tell me what it gets right.

/SCOPE: Before answering, tell me what this question is asking and what it leaves out.

/RECEIPTS: For every claim, mark it sourced, inferred, or guessed. Link sourced claims to my material.

/TRADEOFF: Show what I lose with each option, not only what I gain.

/PLAIN: Rewrite the explanation for someone who has never seen this topic. Define necessary terms as you go.

Five output codes from the second study reel

Copy this

/FLOWCHART: Turn this into a flowchart. Show each step in order, use arrows, and mark every branch.

/NEEDTOKNOW: Pull out only the must-know facts. Use bullets, cut background detail, and add nothing that is not in my notes.

/ONEBYONE: Ask one question at a time. Wait for my answer, tell me whether it is supported by my notes, then ask the next one.

/MULTIPLECHOICE: Write ten multiple-choice questions from my source. Use four options each. After I answer, show the correct answer and explain why each wrong option is wrong.

/TEN: Explain this concept in plain words. Define every hard word, use one everyday comparison, then explain why it matters.

Things to know

ChatGPT can misread tables, skip qualifiers, or write plausible questions that do not match your instructor's assessment. “Question Predictor” creates practice questions. It cannot know the real exam. For medical, legal, or safety-critical learning, use the original source and instructor guidance as the authority.

If something goes wrong

If outside facts appear, repeat the source rule and ask for the exact supporting line.
If the diagram is confusing, ask for a text-only flow with one action per line.
If questions are too easy, supply the expected difficulty and learning objectives.
If the source is a poor scan, paste the relevant text directly.

Final check

Pick three claims from the output. The workflow passes only if all three trace to the supplied source and ChatGPT labels one deliberately missing answer as unsupported.