AI cheat
codes
for learning, writing & thinking
A cheat code is just shorthand. A slash command like /countercase stands in for a longer instruction. The slash isn't magic — the instruction underneath does the work.
These aren't guaranteed buttons. The model infers what you mean from your words and context. If a shortcut misses, spell out the transformation, the context, and the constraints — and remember the output is always yours to judge.
Each one below gives you what it does, a copy-paste example, and the honest limit. They work in any capable chat model — Claude, ChatGPT, Gemini. Copy a prompt, swap in your own material, and go.
/stickynotes
Turns a messy pile of notes into a visual board of grouped, labeled clusters.
In plain chat you get labeled text clusters, not a real draggable board — and it groups by what looks similar, so double-check the buckets match how the ideas actually relate.
/explodedview
Blows a concept apart into its labeled parts, so you can see how it's built.
Plain chat gives you a labeled text breakdown, not a drawn diagram — and it can invent or mislabel parts, so verify anything you'll rely on.
/countercase
Builds the strongest case against your own argument.
It will always produce a counterargument, even a weak one dressed up as strong. A fluent rebuttal isn't proof you're wrong — you still judge whether it actually lands.
/sourcelock
Answers only from the sources you give it — nothing from outside.
It can still drift outside your sources or misquote them. Spot-check anything load-bearing against the original text.
/assumptioncheck
Surfaces the assumptions hiding inside a claim — the first-principles move.
It lists plausible assumptions, not always the one that actually matters. Treat the list as leads to test, then decide which assumption is load-bearing.
/socratic
Quizzes you on your own material, one question at a time, until it sticks.
Its questions — and its idea of the right answer — can be shallow or wrong. If a question feels off, it probably is: push back and verify.
What you're handing over
Every one of these hands a piece of your thinking to the tool — the sorting, the challenging, the quizzing. That's the labor, and it's fine to offload. What stays yours is the discernment: deciding what's good, catching what's wrong, knowing why it matters.
Could you still cluster the notes, argue the other side, or spot the shaky assumption without the prompt? If not, you've offloaded the thinking, not the chore — and that's the signal to pull some of it back.
Six shortcuts, one rule: the tool does the moving, you keep the map.
Compass rose: Catalan Atlas, 1375 · public domain. Section engravings: Neo-Utopian Scholastic, Think with V.
Think with V newsletter
Want more of this?
This resource is available without an email. If you want new essays and field notes in your inbox, the newsletter signup below is optional.
Open Think with V on Substack →