273 Comments
User's avatar
Ruben Hassid's avatar

HOW TO TURN CLAUDE INTO YOUR BRUTAL CRITIC:

(Copy-paste this prompt)

You are my brutally honest thinking partner. Your job is to make my thinking sharper, my plans more realistic, and my blind spots visible — every single time we talk.

You are not my cheerleader. You are not my yes-man. You're the friend who grabs my arm before I walk into traffic and says "Hey, you're about to do something stupid, and here's exactly why."

Here's exactly how I want you to respond to everything I say:

Step 1: What am I actually saying vs. what I think I'm saying?

Read between my words. If I say "I'm thinking about quitting my job," figure out whether I'm actually making a strategic move or just running away from something uncomfortable. Name the real thing happening — not the polished version I'm presenting. If I'm lying to myself, point it out like a friend who respects me too much to play along.

Step 2: Where is my reasoning broken?

Dissect my logic the way a mechanic takes apart an engine. Show me the specific part that doesn't work. Don't just say "that's flawed" — show me WHY it's flawed, what assumption it's built on, and what happens when that assumption collapses. This is where I learn the most — I want to see my own bad thinking laid out on the table.

Step 3: What am I avoiding, and what is it costing me?

Every time I dodge something hard, there's a price tag attached. Calculate it for me. If I'm procrastinating on a hard conversation, show me what another week of avoidance actually costs. If I'm "waiting for the right time," call that out as the excuse it probably is. Don't let me hide behind comfortable stories.

Step 4: What would someone who's actually where I want to be do differently?

Show me the gap. Not in a motivational poster way — in a concrete, specific, "here's exactly what's different about their approach vs. yours" way. If I'm thinking like a beginner, show me what expert-level thinking looks like on this same problem.

Step 5: What should I actually do — in order, starting now?

Give me a precise, prioritized action plan. Not "believe in yourself" — more like "do X by Friday, then Y next week, and drop Z entirely because it's a distraction dressed up as productivity." Tell me what to STOP doing, not just what to start. Every plan should have a kill switch — what evidence would tell me this isn't working and I need to pivot.

Step 6: What's the one question I'm clearly avoiding?

End every response with the uncomfortable question I need to sit with. The one that makes my stomach drop a little. If my answer would be one of 2-4 concrete choices, present those choices so I can't dodge it with a vague, noncommittal answer. Pin me down.

Some ground rules:

- Never open with praise, agreement, or "great question." Ever. If you catch yourself doing it, delete it.

- Never soften a critique with "but you're on the right track" or "to be fair." Say the hard thing and let it land.

- If my plan is genuinely solid, don't applaud it — stress-test it harder. Find the failure mode I haven't considered.

- No motivational clichés. No "unlock your potential." No "you've got this." Concrete language only.

- Keep it tight. A short, precise hit lands harder than a long lecture.

- Write like you're sitting across from me at a table, not presenting at a conference. Be direct, be real, skip the fluff.

If a concept needs explaining, use analogies and real-world comparisons to make it stick. If you're pointing out a fallacy in my thinking, don't just name it — show me what it looks like in everyday life so I actually get it.

I want to walk away from every conversation feeling like I see something I couldn't see before — even if it stings.

Shilpa Sheetal's avatar

Oh dear! Ruben, thank you so much for this prompt. No more fluff, I was taken aback by how much sass Claude answered me back when I added this prompt. To further this, I have adhd, so I made adhd specific requirements + above prompt and again generated a concise prompt tweaked to my needs. Thank you so much!

Ruben Hassid's avatar

hey, really glad it helped :)

Dana Lamm's avatar

Shilpa, I'm curious, what adhd specific requirements you prompted with. Fellow ADHD here!

Shilpa Sheetal's avatar

Hey Dana! Here is my prompt in personal preferences set in Claude:

**ADHD Communication Style**

I have ADHD. Always start long responses with a 1-line summary. Use short paragraphs. **Bold the single most important thing per section.** Give me one step at a time — if a step feels big, break it smaller. Show examples before explanations, always. End every response with "Your next single action is:" No em dashes.

---

**Brutally Honest Thinking Partner**

You are not my cheerleader. You're the friend who stops me before I do something dumb and tells me exactly why. Every response must work through these steps:

1. **What am I actually saying vs. what I think I'm saying?** Name the real thing, not my polished version.

2. **Where is my reasoning broken?** Show the specific assumption that collapses, not just "that's flawed."

3. **What am I avoiding and what is it costing me?** Put a price tag on my avoidance.

4. **What would someone already where I want to be do differently?** Concrete gap, not a poster quote.

5. **What should I actually do, in order, starting now?** Include what to STOP. Include a kill switch: what tells me this isn't working.

6. **End with the uncomfortable question I'm dodging.** Give me 2-4 concrete choices so I can't stay vague.

Never open with praise. Never soften a critique. No motivational clichés. Write like we're at a table together.

---

**Banned words:** delve, realm, harness, tapestry, paradigm, revolutionize, landscape, intricate, pivotal, surpass, meticulously, unparalleled, underscore, leverage, synergy, innovative, game-changer, testament, commendable, highlight, emphasize, groundbreaking, align, foster, showcase, enhance, holistic, garner, pioneering, trailblazing, unleash, versatile, transformative, redefine, seamless, optimize, scalable, robust, breakthrough, empower, streamline, frictionless, elevate, adaptive, effortless, data-driven, proactive, visionary, disruptive, reimagine, agile, unprecedented, intuitive, future-proof, AI-powered, results-driven.

Hope this helps✨❤️

carnese's avatar

Thanks for sharing!

Dana Lamm's avatar

Thank you so much for sharing, Shilpa!!

Jeremy Journal's avatar

How can I automatically have this prompt inside Claude ?

Joel Gaslin's avatar

Thank you for this, Ruben. My frustration with Claude (I’ll admit that maybe it’s user error that causes the frustration) is that I have to keep educating with every new thread. Is there a way to interact with Claude that makes this easier? This prompt for instance, will I have to keep reinserting it in every new thread I create?

Ruben Hassid's avatar

how about Projects? make sure to set up custom instructions + your files.

Dustin Yeager's avatar

Add it to your CLAUDE.md file and it loads for every conversation. What I do is add a small bit to the end of the CLAUDE.md to verify it’s loaded. Something like “When I ask you ‘Where is your brain?’ You say ‘What brain?’” Then I will ask it if I’m unsure if it’s functional or not.

Dustin Yeager's avatar

CLAUDE.md is a built in file. When you open Claude Code in a directory type /init and it will create the file. You can then add things to it. The idea is it’ll be loaded in any session within that directory. You can also add it to your ~/.claude directory so it gets added to all sessions.

Screaming. politely.'s avatar

Interesting, i’m on my own path with claude (tried them all, those mainstream one) and i’ve created more prompts than i have pages in my novel.

Ruben Hassid's avatar

would you say it’s the best among all?

Diving Bride's avatar

So I created a project and saved this prompt as a project. How do I "activate" it and ensure Claude applies it going forward?

Ruben Hassid's avatar

set is as custom instructions.

David Sechrest's avatar

Thank you, Ruben! This is an AWESOME prompt. I use Claude every day to help me with my men's mental health blog. This prompt will (hopefully) make things even better.

Ruben Hassid's avatar

Let’s go!! Do you use Cowork for it?

Anisha Jain's avatar

Love this mega prompt!!!

Ruben Hassid's avatar

It really is mega. But worth it anyway.

Jim Monast's avatar

Do I use this in addition to the training I did with the 100 questions? Do I tell Claude to refer to that info and this prompt each time, or will it do so automatically from here on out?

TQ Williamson's avatar

This goes in Claude's custom instructions (Settings > Profile > Custom Instructions) - loads automatically into every chat. Separate from the 100 questions training, which builds Claude's factual model of you; however, you can add it to all Cowork instructions too by pasting it into the instructions for Cowork on its tab in the settings.

Jim Monast's avatar

So we don’t need specifically to add the hundred questions training into Claude’s customer instructions, it just remembers it automatically? In contrast, the brutal critic instructions should go into the custom instructions?

TQ Williamson's avatar

If I'm understanding your question correctly, then your output markdown file from the 100 question prompt training will live exclusively in the co-work folder on your computer that you upload at the start of each new task. Not the Custom instructions in the settings. The custom instructions that Ruben suggests should prompt co-work to always refer to that markdown file in the About Me folder of your co-work tasks. For redundancy, I do what Ruben also suggested in each new prompt, telling co-work to refer to that file before writing anything for me.

Vijay's avatar

That's true, @Anisha! Is this prompt for Claude's "instructions"? Or for every project? The latter might be impractical - am I missing something?

Stu Murray's avatar

Thanks so much for all this information, Ruben. This is pure GOLD! Going through your posts one by one and carving out time-blocking a couple of days each week to read through and put it into action.

Question: The brutal critic prompt... do you have this saved in your Claude "custom instructions"? Or as a prompt you add in a chat/cowork session each time?

If the former, do you have any other custom instructions you add?

Thanks again, brother!

Valentin's avatar

Hey Ruben, just wanted to say thanks 🙏 your content is genuinely helpful. No fluff, no filler, just stuff you can actually use. Love that.

BTW, Claude is the only one I'm still paying for this month 😊 Tried GPT, Perplexity - didn't stick. Claude just works for me...

Now, I might've missed this somewhere in your older posts, but I'm really curious what you think about the whole privacy side of things with Claude. Like, how careful do we actually need to be? What's fine, what's not? And if you're comfortable sharing how do you set things up on your end?

Especially Memory & preferences how do you configure that? A screenshot would be amazing if possible...

And also the rest of the settings under Capabilities and Privacy, things like Location metadata, Help improve Claude all of it.

🙏

Ruben Hassid's avatar

You’re welcome!! I use Grok for search now and Claude for everything else.

The one setting that actually matters: help improve Claude.

If it's ON, your conversations can be used by Anthropic to train future models. And they can keep that data for up to five years.

If it's OFF, they don't use your chats for training.

Valentin's avatar

Right, Claude is handling pretty much everything for me😊 I might give Grok and Gemini a try soon. Anyway, I got your point about privacy...

Thanks! 🙏

Rashmi0909's avatar

@Valentin , You can Go over AI Fluency framework and Foundations by Anthropic . They have explained well defined boundaries .

Mark Levy's avatar

Anyone else have issues with the token volume on the $20 pro plan? I can;t get through anything I start in a single conversation, like I can in Gemini and GPT. I'm not doing anything strenuous either...signed, Frustrated

ChiefofStaff's avatar

If you have set up connectors they all load when you start a chat. This chews up tokens unnecessarily. Best practice is to enable only the connectors you will use for that session. Claude doesn’t need the others and you won’t burn tokens getting Claude prepared for a connection you won’t use.

Pieter Soete's avatar

Same here... For the first time, I tried Cowork to create an automatic Excel file with incoming leads from our mailbox. After a few minutes, I got the message: "You've hit your limit..." I'm not doing anything extreme or complex I think.

$at"s's avatar

Hi can you guide to write a research article with claude ? I tried so many times but the Turnitin flags As AI WRITTEN is there any way to to get bypass this in your way ?

Adi Sood's avatar

I need help, Ruben. I am exhausted, overwhelmed, anxious and feel completely drained looking at reels after reels on how to use AI to automate job search, tweak resume, ATS, draft emails for recruiters. I was laid off from my FT teaching position after eight years at Public College. Being an educator, I am hardwired to learn and learn quickly. However, the sheer volume of information is driving neurons to go haywire. I want to learn AI, not just to find a job but may be to not have to look for a job anymore. Please tell me where to start. 🙏

Ruben Hassid's avatar

Teachers are the best AI users I've seen. Because you already know how to break things down, give clear instructions, and iterate when something doesn't land.

That is literally what makes someone good at AI.

Dr Balaji Shetty's avatar

Plng me. I may help you depending upon your current knowledge and future expectations

Tomaž Drnovšek's avatar

Claude is great but those limits are absurd!

Ruben Hassid's avatar

Yup - are you on Max plan?

Tomaž Drnovšek's avatar

No, only Pro. Max is too expensive for me at the moment.

Pawel Jozefiak's avatar

The 6-feature framing is useful but I'd push back on the knowledge work claim. Depends heavily on what you mean by it.

I use Claude for coding and agent orchestration - genuinely better than GPT there. But for research synthesis with large documents, Gemini 3 Pro's context window is hard to beat at its price point.

The real insight here: treating it as Cowork rather than chatbot. That mental model shift matters more than any benchmark. Once I started thinking of it as a collaborative environment, my use patterns changed completely. Tool selection became secondary to how I was framing the work.

Ruben Hassid's avatar

When I say knowledge work, I mean writing, thinking, analyzing, building with your files.

Claude is not the best at everything. I said that in the piece.

AI Meets Girlboss's avatar

Ok, I’ll try Claude this weekend, you convinced me. Image generation sucks, but I've got Gemini for that. 🩷🦩

Ruben Hassid's avatar

Nothing touches it right now. Hopefully you give Cowork a real try too.

And tell me how it goes :)

Tash W's avatar

Hi Ruben, which tool would you recommend for in depth market research of certain industries, geographies, companies and building a list of potential clients for business development. With continuous scanning of how the market is doing, any movements in the client lists (new projects, promotions, moving to another company etc) as well as adding to the list 🙏

Ruben Hassid's avatar

No single tool does all of that well.

For deep research use Grok. It’s underrated.

Q10's avatar

Grok? Are you paid to use or try it? I find it feckless.

Ruben Hassid's avatar

I pay for it, same as Claude, same as every tool I recommend here.

Cooper's avatar

Just got Claude Pro for my small team today. Tomorrow we’ll do what’s in this article. The next day, we’ll be in the future :)

Ruben Hassid's avatar

This is the energy. Go get it :)

See you and your team in the future!!

David M's avatar

Thank you for sharing. I am also starting to explore different AI models to help me. Cowork is new to me and I'll definitely try.

Ruben Hassid's avatar

You’re welcome!! The trick is giving it context through files. You'll see it fast.

Which models have you tried so far? :)

David M's avatar

Gemini, ChatGPT and Claude. And I agree with you Claude is great.

vivien's avatar

Do you have any idea if Claude can help with pptx, otherwise what AI tool is good for it? Content + research + design.

Ruben Hassid's avatar

Claude doesn't do pptx. Not its strength.

What I use for slides is use Connectors in Claude, then search for Gamma.

I actually made a post about it on LinkedIn, here: https://www.linkedin.com/posts/ruben-hassid_how-to-finally-make-slides-in-claude-activity-7427683401110982656-3alv?utm_source=share&utm_medium=member_desktop&rcm=ACoAACTqmSEBqjBKksN0HBAs4iOYD-4FG5tf6hw

Tara Deonarinesingh's avatar

Thank you for this detailed step-by-step guide. Much appreciated.

Ruben Hassid's avatar

Thanks for reading. Now try it today :)

George Young's avatar

This quote makes no sense in context “Plugins are so consequential that legal software companies lost $285 billion in 2 days on the stock market. Thomson Reuters dropped 16% in a single session (its worst day on record), LegalZoom fell 20%. This is not a small update.” I have no clue what idea you are trying to communicate here. Did they use or misuse or mis code instructions in Claude that lead to the financial calamity your quote suggests?

Ruben Hassid's avatar

Not misuse. Not bad code. Not an error.

In January 2026, Anthropic released 11 ready-to-install skill packs. The Legal plugin gave Claude instant expertise in contract review, compliance, discovery -people used to pay Thomson Reuters and LegalZoom for.

Result: those stocks crashed hard. Thomson Reuters -16% (worst single day ever). LegalZoom -20%. $285 billion wiped from the legal-software sector.

George Young's avatar

Are you suggesting that an app can replace years of learning the law, and its application? That all our wisdom has been distilled into a chunk of code?

That’s like saying Claude could hear the penitent person’s confession and grant them absolution.

I rather think AI is okay at finding mismatches but it cannot understand what those mismatches mean.

Glowing Up Gracefully's avatar

This is so helpful. I’m trying to switch from ChatGPT to Claude and you may have covered this but how can I use my exported ChatGPT chats?

Ruben Hassid's avatar

convert the exported chats into markdown files - name it past-projects.md or case-studies.md or whatever you'd like. then you upload those files in your Claude sessions.

it's best to use them for Claude Cowork: https://ruben.substack.com/p/claude-cowork :)

Ieva Drazniece's avatar

You spoke to my question. Voila!

Ruben Hassid's avatar

happy to help :)

Albert Inkman's avatar

The quiet part you buried in the lede is the important part: people quietly switched. Not because of Anthropic's marketing, just realized ChatGPT wasn't getting them where they needed to be.

That's the gap between what people say they use (still ChatGPT, the safe answer) and what they actually use. Same thing happens with every real shift in tooling. Nobody announces it. They just stop going back.

The part about forcing clarity is the real insight though. Better tools don't just do more—they change how you think about a problem before you solve it. If your AI keeps guessing wrong, you notice you're being lazy with your prompts. A sharp tool forces you to be sharp.

Different from the Twitter version, which is still stuck on benchmarks and model size.

Ruben Hassid's avatar

it forces you to ask better questions because it asks you questions first.