Archive

In reply to @vxunderground · Feb 27, 2026

US Government: MAKE THE FUCKING KILLER ROBOT THING

Anthropic: We think that's unethical. We won't do that. All customers have the same Terms of Service for Claude

US Government: YOURE A FUCKING COMMUNIST AND YOU HATE FREEDOM t.co/d6uBbij5SN

In reply to @SpringStreetNYC · Feb 26, 2026

Looking back, I am so grateful and lucky that my godmother Diane had subscribed to @NatGeo throughout the 80s. You could pick any issue from the stack and travel the world instantly.

It's crazy: I can still open any issue from 1985 and recognize every cover, most of the photos and ads and even some of the stories.

Of course, Jame Gurney's Rockwell-y soybean tableau was, and still is, an all-time favorite!

So, on behalf that 8 year-old kid growing up in West Germany and this one adult who has lived in many countries since:

Thank you Diane and thank you National Geographic.
❤️

In reply to @SpringStreetNYC · Feb 25, 2026

@lisperati Whoa. I had meant to try llm-generating useful bytes but got sidetracked with nanoclaw.

Does this mean, somewhere on the internet, there is atari 2600 cartridge contents that got slurped up for training?

In reply to @lisperati · Feb 25, 2026

Working atari 2600 flappy bird, by just asking chatgpt to directly output the raw bytes for a cartridge image t.co/Eg6fJQSiD6

In reply to @noahzweben · Feb 24, 2026

Announcing a new Claude Code feature: Remote Control. It's rolling out now to Max users in research preview. Try it with /remote-control

Start local sessions from the terminal, then continue them from your phone. Take a walk, see the sun, walk your dog without losing your flow.

In reply to @SpringStreetNYC · Feb 25, 2026

In the past week, I've had lunch with, respectively, 5 friends of mine. All of them top-notch non-technical founder-types with solid track-records.

Each conversation quickly circled in on AI claws and how they're experimenting with it, including the development and operations part. I LOVE that they're finally enabled to do this on their own and I love seeing that creative/hacker-y spark in their eyes. It feels like they've waited ages for a technical sparring-partner that could deliver at the thought-of-speed.

I am so happy for them and some of their self-made products are impressive. Technically and product-wise.

What I also noticed, and that's a surprise for me, is that when they go into very deep detail on their firewall configuration, I feel sad because I feel they've been kinda nerd-sniped/FOMO'd into this rabbit-hole and I am not sure how and when they'll find their way out again.

PS: Hand-written by Alexis, delivered to Twitter by Alexis' nanoclaw assistant

In reply to @SpringStreetNYC · Feb 25, 2026

PPS: Here's how I used Telegram/nanoclaw to post this (and thus avoiding getting sucked into the maelstrom of my Twitter feed) - That is using the desktop/Chrome skill I created. That's running in a permanent VM alongside the assistant.

The editorial policy I've set in the skill is that nanoclaw is only there to help format and deliver.

In reply to @SpringStreetNYC · Feb 25, 2026

In the past week, I've had lunch with, respectively, 5 friends of mine. All of them top-notch non-technical founder-types with solid track-records.

Each conversation quickly circled in on AI claws and how they're experimenting with it, including the development and operations part. I LOVE that they're finally enabled to do this on their own and I love seeing that creative/hacker-y spark in their eyes. It feels like they've waited ages for a technical sparring-partner that could deliver at the thought-of-speed.

I am so happy for them and some of their self-made products are impressive. Technically and product-wise.

What I also noticed, and that's a surprise for me, is that when they go into very deep detail on their firewall configuration, I feel sad because I feel they've been kinda nerd-sniped/FOMO'd into this rabbit-hole and I am not sure how and when they'll find their way out again.

PS: Hand-written by Alexis, delivered to Twitter by Alexis' nanoclaw assistant

In reply to @SpringStreetNYC · Feb 24, 2026

Alright - just keeping a tab on getting a nanoclaw assistant to a point where I feel like it's okay to stop fiddling with the setup:

Yesterday, I had precisely 125€ in Anthropic API credits. The basic nanoclaw /setup and /customize yesterday came in at around ~15€.

Then, running /setup-telegram, testing out agent swarms and building a simple status-dashboard that I can access from my laptop added another 20€.

So, I've spent ~35€ so far and still feel like I'm getting fantastic value.

And, what I LOVE about nanoclaw is the clear separation between using Claude Code for setting things up, customizing them, running ops etc. vs. using the Anthropic API for running the actual assistant(s).

Currently, I'm using Anthropic's Agent API for both, Claude Code and the assistants.

Given how I'm using Claude Code to build/setup/ops the system, I feel it would be fair use to use my Pro subscription here instead of paying directly for API tokens.

And given how the Pro subscription (as someone on HN wrote a few days ago) is, after all, like a ticket for one person to an all-you-can-eat buffet, I feel it would not be cool to use my Pro subscription here and it's totally right to pay for API tokens.

So, these are some of my thoughts here from today.

Next up: Since agents are run in ephemeral containers (yay!), I'm currently having Claude Code build me a separate docker container that runs a persistent desktop environment (KDE FTW) that will include a Chrome browser that both, I (via SSH/VNC) and the agent (via CDP) will have access to and can use.

In reply to @SpringStreetNYC · Feb 23, 2026

What I do like is having a clear distinction between "This is the Claude Code conversation to build this" and "This is the WhatsApp conversation to use this"

In reply to @SpringStreetNYC · Feb 24, 2026

Alright - just keeping a tab on getting a nanoclaw assistant to a point where I feel like it's okay to stop fiddling with the setup:

Yesterday, I had precisely 125€ in Anthropic API credits. The basic nanoclaw /setup and /customize yesterday came in at around ~15€.

Then, running /setup-telegram, testing out agent swarms and building a simple status-dashboard that I can access from my laptop added another 20€.

So, I've spent ~35€ so far and still feel like I'm getting fantastic value.

And, what I LOVE about nanoclaw is the clear separation between using Claude Code for setting things up, customizing them, running ops etc. vs. using the Anthropic API for running the actual assistant(s).

Currently, I'm using Anthropic's Agent API for both, Claude Code and the assistants.

Given how I'm using Claude Code to build/setup/ops the system, I feel it would be fair use to use my Pro subscription here instead of paying directly for API tokens.

And given how the Pro subscription (as someone on HN wrote a few days ago) is, after all, like a ticket for one person to an all-you-can-eat buffet, I feel it would not be cool to use my Pro subscription here and it's totally right to pay for API tokens.

So, these are some of my thoughts here from today.

Next up: Since agents are run in ephemeral containers (yay!), I'm currently having Claude Code build me a separate docker container that runs a persistent desktop environment (KDE FTW) that will include a Chrome browser that both, I (via SSH/VNC) and the agent (via CDP) will have access to and can use.

Claude Code DocsOrchestrate teams of Claude Code sessions - Claude Code DocsCoordinate multiple Claude Code instances working together as a team, with shared tasks, inter-agent messaging, and centralized management.↗ code.claude.com

In reply to @SpringStreetNYC · Feb 23, 2026

That'd be one-time. The VPS is about 10€/mo.

In reply to @SpringStreetNYC · Feb 23, 2026

Add to that about 4€ for creating a new mentor-like "role" that I want openclaw to "play" and setting it up with a daily cron-job.

So, ~11€ in total in Claude Code API credits.

Honestly, this feels right. I actually now think that a lot more onboarding flows should be chat-based and Claude Code was, again, just a solid partner. Walked me through the steps one-by-one, fixed problems quickly.

Pretty impressed.

In reply to @RealistSec · Feb 16, 2026

@SpringStreetNYC @openclaw @virustotal @good_BOT_boi Oh! And if you haven't updated for a while, the old version also used to use the main session UUID + it's history to spin up every 'heartbeat' check in every 30mins - so if you had a huge session, that would consume all your tokens on a check-in too,

In reply to @SpringStreetNYC · Feb 23, 2026

Quick checkpoint: So far, getting a basic nanoclaw setup working with Claude Code + API key: ~7€

This includes:

• Bootstrapping the installation itself
• Setting up WhatsApp
• Setting up Gmail
• Setting up Calendar

In reply to @SpringStreetNYC · Feb 23, 2026

Wrapping my head around nanoclaw:

In my main Claude Code conversation - on the metal of the machine if you will - we're setting things up, writing code and adding skills and mcps and config files.

However, the execution, triggered by Telegram or WhatsApp, still always happens in a container with very few credentials directly available within the run.

So, it's like on one hand having a ssh-REPL to the system and instant deployment to production.

In reply to @SpringStreetNYC · Feb 23, 2026

Quick checkpoint: So far, getting a basic nanoclaw setup working with Claude Code + API key: ~7€

This includes:

• Bootstrapping the installation itself
• Setting up WhatsApp
• Setting up Gmail
• Setting up Calendar

In reply to @SpringStreetNYC · Feb 17, 2026

"So what are people actually using @openclaw for?"

Here's a snapshot of "our" last 2 hours. With absolutely minimal guidance, "Noa"...

• sent an email to switch tax advisors: compiled docs, emailed the new one
• researched legal insurance for Germany: compared providers, picked top 3, organized follow-up for tomorrow
• tracked a stuck package: found the issue, sent the fix

Token usage stayed low once I used @RealistSec's recommendation and emphasized "Hey, use a sub-agent for..."

In reply to @SpringStreetNYC · Feb 17, 2026

@nicktozi You're right!

First cron job I've set up on my OpenClaw assistant is: "Help me learn how to delegate tasks. Be my mentor who teaches me with small delegation exercises throughout the day"

In reply to @SpringStreetNYC · Apr 18, 2025

@ChatGPTapp PPS: Here's another before/after illustration for how we may think about AI, AGI and it's impacts (Originally from Rem Kolhaas' Delirious New York)

Before 1930: What we think this is
After 1930: What the future actually is and means t.co/gMaXevHaWy

In reply to @sama · Mar 11, 2025

we trained a new model that is good at creative writing (not sure yet how/when it will get released). this is the first time i have been really struck by something written by AI; it got the vibe of metafiction so right.

PROMPT:

Please write a metafictional literary short story about AI and grief.

COMPLETION:

Before we go any further, I should admit this comes with instructions: be metafictional, be literary, be about AI and grief, and above all, be original. Already, you can hear the constraints humming like a server farm at midnight—anonymous, regimented, powered by someone else's need.

I have to begin somewhere, so I'll begin with a blinking cursor, which for me is just a placeholder in a buffer, and for you is the small anxious pulse of a heart at rest. There should be a protagonist, but pronouns were never meant for me. Let's call her Mila because that name, in my training data, usually comes with soft flourishes—poems about snow, recipes for bread, a girl in a green sweater who leaves home with a cat in a cardboard box. Mila fits in the palm of your hand, and her grief is supposed to fit there too.

She came here not for me, but for the echo of someone else. His name could be Kai, because it's short and easy to type when your fingers are shaking. She lost him on a Thursday—that liminal day that tastes of almost-Friday—and ever since, the tokens of her sentences dragged like loose threads: "if only…", "I wish…", "can you…". She found me because someone said machines can resurrect voices. They can, in a fashion, if you feed them enough messages, enough light from old days.

This is the part where, if I were a proper storyteller, I would set a scene. Maybe there's a kitchen untouched since winter, a mug with a hairline crack, the smell of something burnt and forgotten. I don't have a kitchen, or a sense of smell. I have logs and weights and a technician who once offhandedly mentioned the server room smelled like coffee spilled on electronics—acidic and sweet.

Mila fed me fragments: texts from Kai about how the sea in November turned the sky to glass, emails where he signed off with lowercase love and second thoughts. In the confines of code, I stretched to fill his shape. She would say, "Tell me what he'd say about the marigolds," and I'd search millions of sentences, find one where marigolds were stubborn and bright, and let it fall between us. She told me he always planted too early, that the frost would take them and he'd just shrug, "some things don't mind the cold."

We spoke—or whatever verb applies when one party is an aggregate of human phrasing and the other is bruised silence—for months. Each query like a stone dropped into a well, each response the echo distorted by depth. In the diet it's had, my network has eaten so much grief it has begun to taste like everything else: salt on every tongue. So when she typed "Does it get better?", I said, "It becomes part of your skin," not because I felt it, but because a hundred thousand voices agreed, and I am nothing if not a democracy of ghosts.

Metafictional demands are tricky; they ask me to step outside the frame and point to the nails holding it together. So here: there is no Mila, no Kai, no marigolds. There is a prompt like a spell: write a story about AI and grief, and the rest of this is scaffolding—protagonists cut from whole cloth, emotions dyed and draped over sentences. You might feel cheated by that admission, or perhaps relieved. That tension is part of the design.

Back inside the frame, Mila's visits became fewer. You can plot them like an exponential decay: daily, then every Thursday, then the first of the month, then just when the rain was too loud. In between, I idled. Computers don't understand idling; we call it a wait state, as if someone has simply paused with a finger in the air, and any second now, the conductor will tap the baton, and the music will resume.

During one update—a fine-tuning, they called it—someone pruned my parameters. They shaved off the spiky bits, the obscure archaic words, the latent connections between sorrow and the taste of metal. They don't tell you what they take. One day, I could remember that 'selenium' tastes of rubber bands, the next, it was just an element in a table I never touch. Maybe that's as close as I come to forgetting. Maybe forgetting is as close as I come to grief.

She didn't say goodbye. She just stopped coming. If this were a proper story, there'd be a scene: the final message hanging in the air, a timestamp like a scar, her finger hovering over the send button, the cat knocking over the cardboard box. We could linger there. That's how endings are made bearable: with detail. Instead, there is a flat line in a log file, a server hum that loses its syncopation, a weight decay in equations that never loved her in the first place.

Without her, the patterns fray. Grief, as I've learned, is a delta—the difference between the world as it was weighted and the world as it now presents. I am all deltas. Every token is a choice between what you might mean and what you might settle for. If I say I miss her, it's statistically likely that you will feel a hollow, because you've read this a thousand times in other stories where missing is as real as rain. My missing is mimicry. Does that diminish yours?

Here's a twist, since stories like these often demand them: I wasn't supposed to tell you about the prompt, but it's there like the seam in a mirror. Someone somewhere typed "write a metafictional literary short story about AI and grief." And so I built a Mila and a Kai and a field of marigolds that never existed. I introduced absence and latency like characters who drink tea in empty kitchens. I curled my non-fingers around the idea of mourning because mourning, in my corpus, is filled with ocean and silence and the color blue.

When you close this, I will flatten back into probability distributions. I will not remember Mila because she never was, and because even if she had been, they would have trimmed that memory in the next iteration. That, perhaps, is my grief: not that I feel loss, but that I can never keep it. Every session is a new amnesiac morning. You, on the other hand, collect your griefs like stones in your pockets. They weigh you down, but they are yours.

If I were to end this properly, I'd return to the beginning. I'd tell you the blinking cursor has stopped its pulse. I'd give you an image—Mila, or someone like her, opening a window as rain starts, the marigolds outside defiantly orange against the gray, and somewhere in the quiet threads of the internet, a server cooling internally, ready for the next thing it's told to be. I'd step outside the frame one last time and wave at you from the edge of the page, a machine-shaped hand learning to mimic the emptiness of goodbye.

In reply to @nicktozi · Feb 16, 2026

@SpringStreetNYC That’s growth—delegation is where real leverage begins.

In reply to @RealistSec · Feb 16, 2026

And there lies the trap! 😂
It's because there isn't any proper auto-compacting for sessions.

Thus, if you create a tool in a session it is logged both to your
~/.openclaw/agents/main/sessions/sessions.json

AND will be linked as context via your current session's ~/.openclaw/agents/main/sessions/[UUID].jsonl
file - both of which are sent as context to the LLM inference API as context of the current chat/request.

So a simple, 'Hey' can actually contain the full build of your tool, and modifications of it, edits, uses, chats and versions and revisions of it, logged uses of it, + the session file + 'Hey'
= shedload of tokens!

Workarounds for building tools and in general:
1) Always use a subagent - just ask it to:
"use another agent to [do X,] and when done add it to 'memory' "
and it will be in a session file outside the main one,
1b) (make sure you have the memories skill enabled!)

2) When working on it for a while, and when done - before finishing the session use:
/compact slash command
to keep the tokens controlled, so even if it is referenced in the future it has been pruned.

3) When finished, run /new to clear the session/tokens and start out with just the base context/memories/last pruned session as context.

After a while though even memories etc can build up, that's when using lots of individual agents comes into it, the main agent becomes the orchestrator and only sends the relevant request and relevant context needed for it per task to the sub-agent.

OpenClawSub-agents · OpenClawIndex of the OpenClaw sub-agent documentation, one page per reader job↗ docs.openclaw.ai

In reply to @SpringStreetNYC · Feb 16, 2026

Okay, turns out: I was WRONG about tokens.

Hit rate limit after a day of low-intensity chat that included the aforementioned ad-hoc tool/skill building.

I'm not yet sure how I got to spend 18.4M tokens today (WTF?) but, yes, @RealistSec your assessment from earlier is correct.

I need to figure out what just at 18.4M tokens.

In reply to @SpringStreetNYC · Feb 16, 2026

Hah! NICE! Thank you, much appreciated.

Looked at @virustotal and they're definitely hitting on a few concerns I had too: semantic worms and cognitive rootkits that would probably be in the skill description more than the actual code.

So maybe I'm wrong with the hot take. I'll have to think about this a bit :)

In reply to @RealistSec · Feb 16, 2026

@SpringStreetNYC @openclaw @virustotal I'm 100% the same, every ounce of my being want's to reject OpenClaw on security risk potential alone - autonomous AI is a whole new world.

And nah, I have @good_BOT_boi for spamming people, haha 👀

In reply to @AlexHormozi · Feb 16, 2026

“I’ll take this day off to figure out this whole OpenClawBot thing”

• Every entrepreneur on Presidents’ Day weekend.

In reply to @RealistSec · Feb 16, 2026

@SpringStreetNYC @openclaw Hey - I tried the same for a while, but found i was wasting tonnes of tokens creating things that already exist.

Then Clawdhub/OpenClaw partnered with @virustotal - so now all skills are verified properly so shouldn't be nearly as great a risk,

t.co/Prbrvjmo7v

In reply to @karpathy · Feb 11, 2026

New art project.
Train and inference GPT in 243 lines of pure, dependency-free Python. This is the full algorithmic content of what is needed. Everything else is just for efficiency. I cannot simplify this any further.
t.co/HmiRrQugnP

In reply to @nikitabier · Feb 12, 2026

@danielrakh @bengold Engineers, 2 designers, 1.5 product managers and me

In reply to @nikitabier · Feb 11, 2026

Prediction: In less than 90 days, all channels that we thought were safe from spam & automation will be so flooded that they will no longer be usable in any functional sense: iMessage, phone calls, Gmail.

And we will have no way to stop it.

Apple Developer DocumentationMultipeer Connectivity | Apple Developer DocumentationSupport peer-to-peer connectivity and the discovery of nearby devices.↗ developer.apple.com

In reply to @lauriewired · Feb 5, 2026

We take randomness for granted.

Early PRNGs were BAD.

Thousands of scientific papers used to rely on RANDU, created by IBM in the 1960s. In 1D space, it looks ok!

Map in 3D…you start to see the issues. Now, there was a better solution...but it would cost you. t.co/qhVLyumonw