The anti-brainrot Claude prompt pack

I am working through a problem and I want your angles, not your answer. Here is the situation: [describe it honestly, including the constraints you actually have and the thing you would rather not admit].

Give me the four or five genuinely different ways of framing this problem, not four versions of the same frame. For each one, tell me what it assumes, who would naturally see it that way, and what it would push me to do next. Rank them by which you think is most likely to be the frame I am currently missing.

Do not tell me what to do at the end. Instead, ask me which frame feels most uncomfortable, because that is usually the useful one.

Before you give me any opinion on [the decision or problem], interview me. I want to work this out myself with better inputs, not have you decide for me.

Ask me the five sharpest questions that would actually change the answer here. Ask them one at a time and wait for my reply before moving on, and if an answer I give is vague, push back and ask again rather than politely accepting it. Do not soften questions to be nice.

Only after all five, give me your read, clearly labelled as your opinion so I can argue with it, and tell me which of my answers most influenced it.

Here is my plan: [paste it in full, including the timeline and what you are betting on]. Do not improve it for me and do not compliment it.

Find the weakest assumption in it, the thing I have clearly not thought about, and the specific conditions under which this fails badly rather than just underperforms. Then tell me what would have to be true in the world for this plan to work, and rate how confident you are that each of those things is actually true.

Finish by making me defend it: give me your three strongest objections as questions directed at me, and wait for my answers before saying anything else.

I have been offered [paste the offer in full, or the email, with names removed]. Here is my context: [your audience size or experience level, what is being asked of you, how much work it actually is, and what you have been paid before if anything].

Break the offer down into what is actually being bought: the deliverables, the usage rights, the exclusivity, the timeline, and anything that quietly extends beyond the obvious work. Tell me which parts are standard for something at my level and which parts are unusual or one-sided, and explain why for each.

Then give me an honest read on the number: whether it is low, fair or good for this scope, what the range typically looks like, and which single term (not necessarily the price) is worth the most to me if I could only change one thing.

I want to counter this offer without losing it. Here is the offer and my read on it: [paste both]. My relationship with them is [new and cold / warm / they came to me], and my honest walk-away point is [number or condition].

Draft the reply in my voice: warm, direct, no apologising for asking, no long preamble. Anchor on the value I bring rather than on my costs. Ask for the specific change I want, give one short reason for it, and include a fallback I would genuinely accept so the conversation has somewhere to land.

Then, separately from the draft, tell me the two most likely responses they will send back and how I should answer each. Flag any sentence in your draft that weakens my position so I can decide whether to keep it.

Read this document like someone whose job is to protect me, and assume the other side wrote it to favour themselves: [paste the contract, brief or terms].

List every clause that could cost me money, time or ownership later, in plain English, with the exact wording quoted next to each one. Pay particular attention to anything about usage rights, exclusivity, how long it lasts, who owns the work, what happens if either side walks away, and payment timing.

For each item, tell me what a reasonable version of that clause would look like, and rank the whole list by what I should push back on first. Be clear that you are not a lawyer and that anything high-stakes needs a real one, then tell me which specific items on your list you think genuinely warrant that.

I understand [topic A] reasonably well and [topic B] reasonably well, but not what happens where they meet. Explain the intersection.

Show me where the two genuinely interact rather than sit next to each other, where an idea from one breaks or has to be modified in the other, and what someone who knows both can see that someone who knows only one cannot. Use one concrete worked example that requires both.

At the end, tell me the single most common mistake people make when they try to apply one of these to the other, and what to watch for in my own thinking that would signal I am making it.

Here is everything I currently understand about [topic]: [write out your actual mental model, messily, including the parts you are unsure about].

Do not correct me line by line. Instead, tell me what the underlying pattern is that my version is circling but not quite naming, and what the people who are genuinely good at this understand that my model does not include yet. Point at the structure, not the trivia.

Then tell me the one thing in my explanation that is subtly wrong rather than obviously wrong, because that is the one that will keep costing me. Explain why it is wrong and what the accurate version is.

I have just learned about [topic] and I want to find out whether I actually understand it or only recognise it. Test me properly.

Ask me to explain it back to you in my own words, then probe the weak spots in my explanation with follow-up questions rather than telling me the answers. Give me one applied scenario where I have to use the idea rather than describe it, and one case at the edge where the obvious answer is wrong.

When we are done, tell me honestly which parts I understand, which parts I am pattern-matching without understanding, and what I should go and read or try next to close that specific gap.

Build me a skill that changes how you respond when I bring you a decision, a plan or something I am trying to understand, so you make me think instead of thinking for me.

Write the SKILL.md. Give it a short lowercase hyphenated name (do not put "claude" or "anthropic" in the name, or the upload gets rejected) and a description that says clearly when to use it: whenever I ask what I should do, ask for a plan or strategy, or ask you to explain something I am learning. The instructions should tell you to ask me the questions that would change the answer before offering an opinion, to argue against the option I am leaning toward, to label your opinions as opinions so I can push back, and to hand the decision back to me rather than closing it.

Add an explicit exception so this does not get annoying: for straightforward tasks like formatting, drafting boilerplate, or a quick factual lookup, skip all of it and answer directly. Then package it as a zipped folder I can upload.