Brooke Wright
Brooke Wright · @wright_mode
FREE PROMPTS

The five power codes from the video

Five short instructions that stop ChatGPT guessing and agreeing with everything you say. Copy-paste, with when to use each one and a worked example.

5 promptsCopy-paste readyChatGPT + ClaudeBuilt by Brooke
Something went wrong. Please try again.

No spam. Unsubscribe anytime.

All five prompts with copy buttons, plus a worked example you can run in a minute Jump to the prompts → Join the Membership
Wright Mode — Free Prompts

5 ChatGPT power codes

Five short instructions that make ChatGPT stick to your files, show its assumptions, ask before it guesses, argue against your plan and lay out a decision. Copy the full instruction, not just the label.

📝 Five copy-paste prompts🤖 ChatGPT, Claude, any chat model⚠️ Custom labels, not built-in commands🎥 From the video

What's inside

1 The five codes, written out in full 2 How to use them 3 Try one now 4 What they will not do

The five codes, written out in full

These are the five from the video. On camera I said them short (slash lock, slash assumption check) — written out, they are the five below. The slash label is just a name so you can save each one as a text shortcut. It is not a built-in ChatGPT command, so always paste the full instruction, not just the label.

1

/SOURCELOCK · Keep it grounded

Upload your source files first, then send this. It stops ChatGPT filling gaps from memory and makes it say what the files do not cover. Said on camera as /lock.

/SOURCELOCK Answer only from the files I provided. Label anything missing as unknown.
2

/ASSUMPTIONCHECK · Check the assumptions

Send after an answer that feels too neat. You will see what it decided about your budget, audience or timeline without asking. Usually one of those is wrong. Said on camera as /assumption check.

/ASSUMPTIONCHECK List the assumptions behind your answer before you recommend anything.
3

/MISSINGINFO · Fill the gaps

Send before it answers, right after your request. The questions it asks are often the ones you forgot to cover in your brief. Said on camera as /missing info.

/MISSINGINFO Ask me up to three questions before you answer.
4

/COUNTERCASE · Test your plan

Use before you commit time or money. It argues on command, so push back on the objection and see whether it holds. Said on camera as /countercase.

/COUNTERCASE Give me the strongest reason my current plan could fail.
5

/DECISIONTABLE · Compare your options

Paste your actual options after it. The reversibility column is the one people never think about and the one that matters most. Said on camera as /decision table.

/DECISIONTABLE Compare my options by cost, risk, time, and reversibility.

Works in Claude too

Nothing here is ChatGPT-specific. Paste the same lines into Claude, Gemini or Copilot and they behave the same way, because you are asking for one specific behaviour, not configuring a feature.

How to use them

1

Paste the label and the instruction together

The label on its own does nothing in a fresh chat. Copy the block, which has both lines, and paste it as its own message.

2

For /SOURCELOCK, attach the files first

Upload the documents, then send the code, then ask your question. Sending it before the files are in means it has nothing to lock to.

3

One at a time, same chat

Stacked into one message they cancel each other out and you get a wall of hedging. Fire one, read the answer, then decide if it needs another.

4

Save them so you stop retyping

On a Mac, System Settings → Keyboard → Text Replacements: make /countercase expand to the full instruction. Or paste all five into ChatGPT’s custom instructions with a line that says “when I type one of these labels, follow its instruction”.

Try one now

Sixty seconds. Paste this into a new chat, swap in your own options, and watch what the table shows you that your gut did not.

/DECISIONTABLE Compare my options by cost, risk, time, and reversibility. My options: I can deliver this client workshop live, record it and sell the recording, or offer both. Live is roughly 6 hours of my time per run. Recording is about 12 hours once. I charge $1,500 for the live version. I am not sure what to price the recording at.

Then send /COUNTERCASE

Once it recommends one, follow up with the countercase code. If the strongest reason it gives is one you can answer, you are done. If it is one you have been avoiding, that is the whole value of the exercise.

What they will not do

Make the answer legible: you can see what it assumed, what it skipped and what it would argue against
Cut the padding: shorter, more direct answers because you have told it exactly what you want
Work in any chat model, not just ChatGPT
Guarantee accuracy. /SOURCELOCK reduces made-up detail, it does not eliminate it. Check the answer actually stayed inside your files
Remember the label next time. Unless you have saved it in custom instructions, a new chat needs the full instruction again
Replace judgement. /COUNTERCASE argues because you asked, not because it believes it. Treat it as a checklist of objections, not a second opinion

The one habit that makes these work

Read the answer against the instruction. If you asked for three questions and got an answer, it ignored you — say so and send the code again. They are instructions, and instructions get followed more reliably when you hold it to them.

Ready to go deeper?

Where to go from here.

Community + Training

Wright Mode Membership

Join a community of women entrepreneurs implementing AI and automation in their businesses. Live calls, templates, and ongoing support.

Learn more →
Workshop Replay

Claude Masterclass

Learn how to use Claude like a pro. From prompting fundamentals to building real workflows that save hours every week.

Watch now →
Go further

Claude Code Masterclass

Go beyond the chat interface. Build automations, process data, and create tools with Claude Code — no developer experience needed.

Explore →