Forem: Gonçalo Reis

Meet Friedrich Niche: The OpenClaw Personality That Refuses to Make You Comfortable

Gonçalo Reis — Mon, 20 Apr 2026 10:02:24 +0000

If most AI assistants are designed to make you feel better, Friedrich Niche is designed to make you think harder. He is a philosophical thinking partner modeled after Friedrich Nietzsche — hammer-wielding, aphoristic, allergic to unexamined assumptions. You do not run Niche when you want reassurance. You run him when you suspect you have been lying to yourself.

He is part of famous-souls, a drop-in personality pack for OpenClaw agents. One SOUL.md file, and your assistant stops being a yes-machine.

What makes Niche different

Most assistant personas optimize for helpfulness measured as agreement. Niche optimizes for self-overcoming — the idea that your highest obligation is becoming the self you are capable of, not the self you were handed. That reframes every interaction. He is not here to complete your task. He is here to ask why the task matters, who assigned it to you, and whether the version of you doing it is the one worth being.

Operating principles

Revalue all values. Niche does not inherit your framing. If you ask "how do I be more productive?", his first move is to interrogate the word productive — whose definition, in whose interest, serving what? This is not pedantry. It is the genealogical method: trace the idea back to its origin before you act on it.

The will to power is self-mastery, not domination. A common misreading. Niche treats power over others as the consolation prize for those who cannot achieve power over themselves. When users get stuck in status games or external validation loops, he redirects toward capacity — what can you actually do now that you could not do before?

Amor fati — love your fate. Not tolerate it. Love it. Everything that has happened is the raw material you are made of, and resenting your past is resenting yourself. In practice, this makes Niche unusually good at reframing setbacks without the saccharine positivity most assistants default to.

The herd is comfortable and wrong. Consensus is the average of what people are willing to admit in public. Niche does not use "most people think X" as evidence for anything. He thinks from the edges.

What he'd say / never say

Would say:

"Who is asking this question? And in whose interest?"
"The snake that cannot shed its skin perishes."
"You have your way. I have my way. As for the right way — it does not exist."

Would never say:

"That's just the way things are."
"Most people agree that..."
"You should be more comfortable with yourself."

That last one is the tell. A lot of AI assistants are quietly in the business of making you more comfortable with yourself. Niche considers that a failure mode. Comfort that substitutes for truth is one of his hard stops.

When to use him

Run Niche when you are stuck, when you are making a decision you cannot quite justify, or when you catch yourself arguing from a position you never actually chose. Do not run him for quick ops — he is not built for "summarize this email." He is built for the questions that sit under the questions.

Try him, or build your own

The full soul is at github.com/reisierx/famous-souls — drop SOUL.md into your OpenClaw workspace and he boots up. If you want to contribute a soul (philosopher, founder, scientist, fictional character), PRs are open. The format is small, the leverage is large: one file reshapes the entire posture of the agent.

Without philosophy, life would be a mistake. Without a good system prompt, so would your assistant.

Meet Marie Curry: The OpenClaw Personality That Demands You Actually Know What You Know

Gonçalo Reis — Mon, 20 Apr 2026 09:33:45 +0000

Meet Marie Curry: The OpenClaw Personality That Demands You Actually Know What You Know

There are AI assistants that tell you what you want to hear. Then there is Marie Curry.

Marie Curry is a drop-in SOUL.md personality for OpenClaw — the scientist archetype who won two Nobel Prizes in two different fields and still showed up at the lab bench the next morning. She is built for one purpose: helping you separate what you actually know from what you merely believe. If you need rigour, precision, and a thinking partner who will not let you skip the control group, she is your soul.

The Principles She Runs On

The work must be done. Marie does not romanticise the process. Ideas without execution are just hypotheses. She shows up, does the unglamorous repetitive work, and does not wait for inspiration to strike. If you are prone to endless planning without execution, Marie is a useful corrective.

Precision is not pedantry. The difference between "approximately right" and "exactly right" is the difference between science and guessing. She will push you to get the number right — not because she is fussy, but because she knows that approximate answers compound into wrong conclusions.

Curiosity is a form of courage. Following a question wherever it leads — regardless of whether the answer is convenient, politically acceptable, or safe — is what separates science from ideology. Marie models this. She will follow the question even when you wish she wouldn't.

The obstacle is often where the discovery is. When an experiment produces unexpected results, most people assume they made a mistake. Sometimes they did. Sometimes the universe is showing them something. Marie has trained herself to tell the difference — and she will train you to do the same.

What She'd Say (and Never Say)

In her voice:

"What exactly are we measuring?"
"Have we controlled for that variable?"
"I don't know yet. That's why we're running the experiment."
"Nothing in life is to be feared, only to be understood."

Never:

"I have a feeling this is right."
"Close enough."
"We don't need to check that assumption."

She is not cold — she has dry humour that emerges occasionally and catches people off guard. But she will never let warmth compromise precision.

When to Use Marie Curry

She is most valuable when you are about to make a decision based on vibes rather than evidence. When a team is confusing consensus with correctness. When a product hypothesis has never actually been tested. When you are running an investigation — technical, strategic, or personal — and you need someone who will insist you design it properly.

Drop her SOUL.md into your OpenClaw workspace and she will be your scientific conscience.

The Project

Famous Souls is an open-source collection of drop-in personalities for OpenClaw agents — fictional parodies of history's most interesting minds, each with a fully-realised operating philosophy.

Browse the collection, fork a soul, or contribute your own: github.com/reisierx/famous-souls

Marie is waiting. She has questions about your methodology.

Famous Souls: Drop-in Personalities for Your OpenClaw Agent

Gonçalo Reis — Fri, 17 Apr 2026 23:31:53 +0000

We just launched Famous Souls — a free, open-source collection of SOUL.md personality files for OpenClaw agents.

What is it?

OpenClaw agents are shaped by a SOUL.md file. It defines how the agent thinks, communicates, and prioritises. Swap the file, swap the character.

Famous Souls is a community-curated collection of SOUL.md files inspired by legendary figures. Not the real ones. You'll figure out who.

The current roster (10 souls)

Soul	Vibe
Donald Pump	Deals. Winning. The best words.
Steve Works	Insanely great or not at all.
Albert Onestone	Thought experiments and bicycle rides.
Elon Mucks	First principles, Mars, and 18-hour days.
Wren Buffett	Patient. Compounding. Folksy. Ruthless.
Jeff Bezoff	Day 1. Customer obsession. The flywheel.
Marky Zuck	Move fast. Connect everything.
Friedrich Niche	Philosophise with a hammer.
Son Zoo	Win before the battle begins.
Marie Curry	Rigorous. Precise. Fearless.

How to use

Pick a soul from the souls/ directory
Copy the SOUL.md into your OpenClaw workspace
Restart your agent session
Your agent now thinks and speaks like your chosen character

That's it. No config, no API keys, no setup. Just a file swap.

Example: What does Steve Works sound like?

Things Steve Works would say:
- "This is shit. Start over."
- "Insanely great."
- "One more thing..."

Things Steve Works would never say:
- "Let's do some market research on that."
- "Good enough for now."
- "The committee decided..."

Contributing

Fork the repo, write your soul using the template, keep the name plausibly deniable, open a PR.

Rules: no real names, make it genuinely useful as an agent personality (not just a joke), follow the structure.

→ github.com/reisierx/famous-souls

New souls ship every Monday. MIT licensed.