Are there good human-curated skill repos for design, coding, product, etc that's not "here's 100 skills you separate the slop yourself"
Sergio Duran
Thanks @garrytan! 🧠 Spent the last 48h diving deep into GBrain looking for a serious personal-memory system for my agentic stack. A few days ago I came across your profile and your skill, and I'm in.
A small contribution incoming 👇
Mayor Matt Mahan
There is no excuse for political violence. Ever. I condemn this unequivocally — and for the sake of our democracy and our decency, I hope we all do. Grateful that everyone is safe.
If you want to know what the next generation expects from AI just use Codex with a 7 year old.
"What do you mean it can't build a pet dragon raising game instantly? I'm bored"
ChatGPT for Clinicians unlocking new use cases:
Dr Danish: I just tried Agent Mode with ChatGPT for Clinicians. This is unbelievable.
I might make a video of this… wild.
ChatGPT for Clinicians unlocking new use cases:
Dr Danish: I just tried Agent Mode with ChatGPT for Clinicians. This is unbelievable.
I might make a video of this… wild.
Luther Lowe
So proud of @weijia. She was on the stage less than an hour ago presiding over the abrupt end of the dinner and now she’s in the front row of the White House briefing room waiting to for the President to speak.
Sebastien Bubeck
Re @NandoDF @sama @gabeeegoooh That's exactly my view too, roughly compresses 2 weeks of our previous jobs to 1 day
I’m safe. It was a crazy night.
Lucas Matheus: Just saw in Instagram that @garrytan was at the same First Amendment celebration where the shooting happened and Trump was rushed off stage...
Hope everything's alright there, bro 🫡🙏
Wild day
Laura Fingal-Surma 🚡 frontier urbanist
.@jansramek in the new YC zine:
"San Francisco’s new generation of 'real world' hardware companies, that build drones, energy generation, consumer hardware, medical devices, or anything else, will never be able to build those devices at scale inside the 47 square miles of San Francisco. There simply isn’t enough land or workers available there. To allow those companies to flourish, we have to build a regional ecosystem where R&D can happen in San Francisco, and scaled up factories can be built within 1-2 hours drive, so that R&D staff based in SF or South Bay can get to the factory in the morning, do a day’s of work with the factory staff, and come back home in time for dinner with their kids."
We should have broken ground on California Forever two years ago. The next best time is now.
Liz4SF: if you haven't heard, YC launched a zine machine "where san francisco goes, so goes the world" w/ determination and deliberation that cant be ignored
https://thevoicesf.org/y-combinator-launches-its-zine-machine
I will admit it's down on Github because now many of my PRs begin on OpenClaw
Aufa Wibowo🍉: @garrytan github commit activity
2000s: every company is an internet company
2010s: every company is a software company
2020s: every company is an AI company
2025+: every company is a cybersecurity company
Heidi Hatch KUTV
CBS News senior White House correspondent Weijia Jiang, who is also the president of the White House Correspondents' Association, was sitting next to President Trump as the incident unfolded.
"What was going through my mind is my 7-year-old daughter was there. My husband was there. My parents were there," Jiang said. "On a night where we all came together to celebrate the freedoms and the First Amendment, we also have to think about how fragile they are in this country because … shootings and would-be shootings happen every day."
"And it doesn't matter if it's the White House Correspondents' Dinner or anywhere else in this country. Nobody should have to feel that way. Nobody should have to feel scared to be anywhere in a public place," she added.
Jiang said the annual press dinner is about acknowledging how vital the First Amendment is to democracy, and a chance for the press and the president "to get together in a different context and recognize the important relationship, despite how complicated it might be."
https://www.cbsnews.com/live-updates/trump-evacuated-white-house-correspondents-dinner-security-incident/#post-update-b3949d85
Releaseing GBrain v0.22 - lots of fixes to search and retireval and a new eval system (gbrain-evals is now a separate repo to prevent checkout bloat)
Daniel Jeffries
Never let critical thinking get in the way of extremist fantasy stories.
If you're on either side of this slider, congratulations, you may be a child in an adult costume body!
Black and white thinking is for children.
When you become an adult it's time to put away childish things.
Ethan Mollick:
What I'm learning from flight simulators is that it would be a bit boring to be an amateur cessna pilot but a lot of fun to be an amateur fighter jet pilot.
Vox
more and more i think everyone should start building their own openclaw now.
last wave a lot of people around me dropped out because claude wouldn't let you run it on a subscription. this wave they're coming back because gpt 5.5 with a bit of tuning runs plenty smooth.
which tells you a big part of what matters in this category is the model you plug in underneath. models getting stronger is inevitable. the fact that people are coming back also says the shape itself is right.
two inevitabilities colliding. hard to picture how big an edge your own agent stack gives you one or two tiers from now.
build one before it fully matures, then grow through the bumps. it's the most worth-it thing of the next two years.
what's the biggest risk you see here?
Francois Chaubard
if you want to know how to build a $60B product in 3 years.. this is it. and its yc mentality through and through:
1) ship it.. even half finished. measure market response. and learn.
2) for everything that works expect 10 things that didnt before it
3) be a user of what you make
4) do not be dissilusioned by the trough of dissilusionment. cursor had a bump then very high churn after launch. thought of pivoting to only bug fixing, etc, but because of (3) they kept building what they wanted, iterating to make it good, until it was good.
--
everyone thinks they are smarter than these rules. dont be.
https://youtube.com/shorts/SAtiFmc4nJU?si=RHDvUIyrGP-iWFyn
Oussama Sekkat
I’ve seen people on X dunking on folks like @garrytan @doodlestein and others for sharing SKILL dot md files they've built. They are dismissing these files as "just a markdown file.”
I think this misses the point entirely and I'll try to address that here. Quick thread:
A bad skill file is just text, sure.
A good skill file is compressed expertise, packaged in a format an agent can actually use.
The value is not just in the “markdown file.” The value is the interaction between:
a huge neural network with latent capabilities
a precise, reusable, agent-readable procedure that steers those capabilities toward a specific outcome
That combination is the product.
Saying “it’s just markdown” is like saying Hamlet is “just ink on paper,” or Einstein’s relativity paper was “just a text.”
Technically true. Intellectually useless.
The medium is simple. The content is what matters. And more importantly, the effect of that content on the reader is what matters.
With humans, a book, a coach, a lecture, or painting can change how someone thinks and acts.
With LLMs, text is also the control surface. These models were trained on text, reason through text, call tools through text, and follow procedures through text.
So yes, the skill is “just text.”
But it is text designed to be read by an enormous neural net.
That matters.
A good skill is agent-ergonomic. It does not merely say “do this better.” It encodes workflow, constraints, examples, edge cases, tool usage, failure modes, and success criteria in a way the agent can reliably execute.
That is very different from a casual prompt.
A prompt is often a one-off request.
A skill can be reused, versioned, tested, improved, shared, and loaded at the exact moment an agent needs it.
That turns “vibes-based prompting” into something closer to operational knowledge.
Another way to think about it:
We have built these massive models, but much of their power is latent. Different people can extract very different levels of performance from the same model.
A good skill is a way to actualize a specific slice of that latent capability.
A refactoring skill.
A research skill.
A legal review skill.
A math explanation skill.
A codebase-navigation skill.
Each one can make the same model behave very differently.
I think of Cus D’Amato and Mike Tyson.
Tyson had enormous latent potential. But Cus gave him a system, a style, a discipline, a way to channel that potential.
That’s what good skills are for agents.
They are not magic. They are not all equally valuable. Many will be mediocre or useless.
But dismissing them right off the batt because they are “just markdown” shows a misunderstanding of what LLMs are.
Text is how we trained these systems. (for the most part)
Text is how we steer them.
Text is how we unlock parts of what they can do.
The question is not whether a skill file is “just text.”
The question is whether the text reliably makes the model perform better at a valuable task.
If yes, then it is not “just markdown.”
It is leverage.
Philo Groves
Before the agents became reliably autonomous, we chatted with them in a terminal.
Their context could only fit a million tokens (max). They were slow. They had “hallucinations”. We inspected their every thought. We watched their every output.
The mid-20s were weird for us all.
America needs to go much harder on open source models
Matthew Berman: http://x.com/i/article/2047838665834131456
morluto
Re @garrytan the big reason chinese labs are so willing to opensource is because china doesnt have a saas industry or much of a services economy
chinese labs can't monetize the same way with enterprise deals, so their main business model has been to open source and sell tokens abroad
Shooting oneself in the foot?
Nope.
Shooting oneself in the prefrontal cortex.
News from Science: U.S. President Donald Trump has fired all 24 members of the National Science Board, the body that oversees the National Science Foundation. Many science advocates see it as the latest step by his administration to erode—some would say destroy—the independence of the 76-year-old
Lisan al Gaib
holy IQ mog
Lisan al Gaib: two facts
- Opus 4.7 is a decent upgrade. if it's worse for you it's a skill issue
- GPT-5.5 will absolutely IQ mog Opus 4.7
I’m adding “interestingness” and flexible ontology to GBrain soon so I think it will be useful to be able to configure more types of data and new customizable weights.
Douwe Tjerkstra: Connecting my WHOOP data to gbrain on my claw could be really interesting.
It also feels like a key piece for a complete 360° view of my data.
GBrain is my open source attempt to solve this for myself as I use OpenClaw/Hermes Agent in my own work and fun
I am learning a lot being at the edge and hope it is useful to others
Rohit: Graph is the final boss of memory.
stacking markdown files isn't memory. it's context you keep re-loading into the prompt.
real memory is a graph. nodes, embeddings, traversal. it's how production agents remember.
the article is the workaround. the lecture is the architecture
Peter Yang
"I shipped 9 failed products before one took off...now I'm doing $1M+/month."
Here's my new episode with @tibo_maker, a solo founder who bootstrapped 5 AI products to $1M+ / month.
Tibo walked me through his exact playbook:
✅ How to validate ideas and fail fast
✅ Why his top acquisition channel is still SEO
✅ The pricing sweet spot for AI products
Some quotes from Tibo:
"When people twist your product into something else, that's a very strong signal you have to follow."
"It's easy to lie to yourself [with free users], but if there's no stickiness in the revenue, it's very hard to build a successful business."
"I'm convinced right now that just one person can do the job of 20 people."
📌 Watch now: https://youtu.be/0UnZnonMN9o
Thanks to our sponsors:
@WisprFlow: Don't type, just speak https://ref.wisprflow.ai/peteryang
@linear: The AI agent platform for modern teams https://linear.app/behind-the-craft
Fin Gómez
We’re so proud of our @weijia — showing grace, poise, and calm strength under just terrible circumstances. Grateful she’s our @whca president, grateful for her leadership, especially for those of us in the confusion and chaos of the ballroom last night. She is outstanding.
Ed O'Keefe: WATCH: From our @cbsnews Special Report, @weijia describes what she saw from the front of the ballroom alongside President Trump @POTUS - and what we didn't see behind the scenes.
"post-AGI, no one is going to work and the economy is going to collapse"
"i am switching to polyphasic sleep because GPT-5.5 in codex is so good that i can't afford to be sleeping for such long stretches and miss out on working"
feels like a good time to seriously rethink how operating systems and user interfaces are designed
(also the internet; there should be a protocol that is equally usable by people and agents)
Front row at the White House Correspondents’ Dinner last night, as a guest of WHCA President Weijia Jiang.
A night celebrating the First Amendment turned surreal when Secret Service scrambled to evacuate POTUS, the VP, and cabinet members, overturning chairs in front of 2,000 reporters just moments after I snapped this photo.
Through it all, Weijia led with grace. Proud to call her a friend.
Moments like these remind us how our country must heal. We need to come together in community and re-embrace what actually unites us: life, liberty, and the pursuit of happiness.
Daniel Jeffries
Existential risk mongers are a small, very vocal cult with a lot of very clever online astroturfing skills.
Politicians never waste a good fake crisis, which is why they're perfect for Bernie to try to seize the means of production back from the private sphere.
Imagine if a mainstream politician put the Heaven's Gate suicide cult on the platform in 1997?
That's where we are in American politics at the edges today.
We've jumped the shark into the absurd and we've completely lost the thread of reality in America today.
And it isn't AI that caused this madness. It is good old fashioned human hallucination, story telling and delusion.
Society should never make policy about imaginary future threats that have no actual basis in reality. AI is already used by billions and it is nothing like what people imagined it to be and we're all still here and the problems have been ridiculously minimal even by the standards of past tech revolutions.
If I told you there was a technology where 1.5 million people would die every year and injure 50 million would you sign up for that tech? The answer is yes because it's cars. You probably got in one today and accepted that risk and never even thought about it.
In contrast, the amount of people who died or were injured by AI is a tiny fraction of a percent and even those cases are probably specious at best.
In fact, AI has actually saved countless lives. It's saved lives through drug discovery by helping create the COVID and flu vaccines. It's saved lives through anti-lock breaks. It's saved lives with self driving cars that are 10X safer than human drivers and yet the only thing you read about self driving cars in the news are the incredibly rare stories of the very few times it did actually kill a dog or injure someone.
If that sounds irrational and insane that's because it is.
We're worried about a technology with a nearly perfect track record of safety and are totally fine with a technology that kills people every day?
Welcome to the irrationality of the human mind!
If the pause AI cult had their way we'd still be stuck on GPT2 which was a hallucinating mess and we'd have made no progress to the much more sound and highly aligned and intelligent and useful models we have now.
That's because you can't make progress by imagining it in your mind or writing idiotic essays on Less Wrong to other mental masturbators.
You solve problems in the real world through trial and error.
The way to make AI safe is by building it and iterating in the real world.
There is no other way.
Jay Van Bavel, PhD: What is the one thing that most worries you about AI?
In a survey of 3,700 AI researchers, only 3% of respondents replied existential risk — despite “the prominence given to these risks in media. Far more researchers are worried aout malicious use, misinformation, job losses,
Peter Yang
A great personal agent should:
1. Get work done across email, calendar, Google Workspace, or any API/MCP it's hooked up to
2. Act proactively and reliably (e.g., cron jobs, triggers, follow-ups)
3. Have excellent memory that helps it "just get you" over time
4. Work across web and mobile without slash commands or manual setup
5. Let you switch between text, voice, video, and live calling mid-conversation
6. Be reachable from any 3rd party messaging app, just like a real person
7. Have a personality that makes it fun to talk to
OpenClaw, Claude Code, Codex - the truth is that none of them check all these boxes yet.
Daniel Jeffries
What if I told you there was a technology where 1.5 million people would die every year and injure 50 million would you sign up for that tech?
Hell no, right?
But the answer is actually "hell yes" because it's cars.
You probably got in one today and never even thought it could kill you. You normalized the risk so easily.
In contrast, the amount of people who died or were injured by AI is a tiny fraction of a percent and even those cases are probably specious at best.
In fact, AI has actually saved countless lives.
It's saved lives through drug discovery by helping create the COVID and flu vaccines. It's saved lives through anti-lock breaks. It's saved lives with self driving cars that are 10X safer than human drivers and yet the only thing you read about self driving cars in the news are the incredibly rare stories of the very few times it did actually kill a dog or injure someone.
If that sounds irrational and insane that's because it is.
We're worried about a technology with a nearly perfect track record of safety and are totally fine with a technology that kills people every day?
Welcome to the irrationality of the human mind!
Superintelligent machines can't come fast enough because the real problem is not intelligence in the world.
It's the utter lack of it.
White House Correspondents Insider
“It all happened so fast.” @CBSNews’ @weijia Jiang’s inside account as law enforcement rushed the stage during the White House Correspondents’ Dinner.
Through it all, Jiang stayed steady & poised, a personification of professionalism in a moment no one expects to face.
Rep. Brian Fitzpatrick 🇺🇸
Re @WHCA President and CBS News Senior White House Correspondent @weijia Jiang showed America last night why she is so respected and so admired, not just as a journalist, but as a human being. Weijia worked so incredibly hard to put this event together, to honor and to celebrate the First Amendment and the Freedom of the Press. In the midst of all of the chaos and danger, Weijia conducted herself with class, with poise, with calm and with the utmost professionalism. And in true Weijia form, she immediately traveled to the White House to cover the President’s press conference. Weijia is a true professional and, more importantly, an amazing human being. America loves you, Weijia 🇺🇸
CheesemonkeySF
There is a Brown Act violation in SFUSD's agenda item regarding approval of Ethnic Studies.
Their “Recommended Action” on the agenda lacks any reference to taking action on Ethnic Studies — only on History/Social Studies.
AGENDA:
https://go.boarddocs.com/ca/sfusd/Board.nsf/goto?open&id=DQ9Q3T64C357
CAROL KOCIVAR BLOG:
https://carolkocivar.substack.com/p/sfusd-everyone-supports-ethnic-studies?r=1v8wim&utm_campaign=post&utm_medium=web&triedRedirect=true
PROBLEM QUOTE (Line 11 in the agenda):
*Recommended Action*
The Board approves the Instructional Materials Adoption of Elementary and High School History/Social Studies Curriculum
Fixing this the day before the meeting would not be adequate.
GPT Image 2 can generate diverse images even for detailed prompts
Jeffrey Emanuel: I was curious how much the new ChatGPT image model would vary in its outputs given the same detailed prompt to make a math explainer infographic.
The result: quite a bit! If it’s something important to you, try generating it a couple times, even if the first one looks great.
GPT Image 2 can generate diverse images even for detailed prompts
Jeffrey Emanuel: I was curious how much the new ChatGPT image model would vary in its outputs given the same detailed prompt to make a math explainer infographic.
The result: quite a bit! If it’s something important to you, try generating it a couple times, even if the first one looks great.
Gianl1974
Trump just fired all 24 members of the National Science Board. Every single one. By email. No warning. No reason given. The board has existed since 1950.
The National Science Board is the independent body that oversees the National Science Foundation, the agency that distributes $9 billion in research grants every year.
Its members are scientists and engineers from universities and industry. They serve six-year staggered terms specifically so they cross presidential administrations and stay independent of whoever is in power.
On Friday, every single one of them got the same boilerplate email from Mary Sprowls of the Presidential Personnel Office: "On behalf of President Donald J Trump, I am writing to inform you that your position as a member of the National Science Board is terminated, effective immediately. Thank you for your service."
That's it. That's the whole letter. For 76 years of institutional independence.
The NSF funds the basic science behind MRIs. Cellphones. LASIK eye surgery. GPS. The internet itself. The Antarctic research stations. The deep-space telescopes. The research vessels mapping the ocean floor. Every breakthrough that made America the world's leader in science for the better part of a century traces back through grants this agency made and this board approved.
The board chair, Victor McCrary, was actively advising Congress on Trump's proposed 55% cut to NSF's budget. The board was helping fight back. So Trump fired the board.
Marvi Matos Rodriguez, one of the fired members, told reporters she had been reviewing an 80-page report as part of her board duties just days before being terminated.
Keivan Stassun, a physicist at Vanderbilt, said NSF's leadership had already stopped responding to board oversight requests months ago. "We would ask them, 'Are you following board governance directives?' And their answer would be, in effect, 'We don't listen to you anymore.'"
Now there's no board to answer to.
Rep. Zoe Lofgren of California, the top Democrat on the House Science Committee, called it "the latest stupid move made by a president who continues to harm science and American innovation. Will the president fill the NSB with MAGA loyalists who won't stand up to him as he hands over our leadership in science to our adversaries?"
That's the actual question.
Because while Trump is firing American scientists, China is building research universities at a rate we cannot match. The CDC just buried a study showing vaccines work.
RFK Jr. runs HHS. The EPA is gutted. The Forest Service is being broken. Half of American children are breathing dangerous air. And now the people who decide what gets researched in the United States have all been fired by email on a Friday afternoon.
Theo - t3.gg
It is genuinely insane that Anthropic will bill you differently if you mention certain words in your prompt or have certain files in your codebase
Om Patel: THIS GUY LOST $200 IN ONE DAY BECAUSE THE STRING "HERMES.md" WAS IN HIS GIT COMMITS
HERMES.md is a real convention used in AI agent projects. it's a system prompt specification file. not some obscure edge case
he's on claude max 20x at $200 a month. yesterday claude code hit
Tony Dang
I believe secrets management will be a core agent infrastructure component layer.
We are still aligning on the optimal form factor for it but @infisical will be the winner - the fully vertically integrated security infrastructure platform that’s actually building toward what agents need.
Very excited for Agent Vault. We will deliver this open source and create the absolute best form factor for all agents.
There’s a lot of work ahead!
Jonhnson Nakano: @dangtony98 This framing makes a lot of sense.
I think agent infra is splitting into a few boring but critical layers:
secrets: what agents are allowed to access
tools: what agents are allowed to call
memory/state: what agents are allowed to remember
artifacts: what agents produce and need
Dima
My @Replit agent worked for a full hour and 3 minutes yesterday with one prompt and the results are impressive too.
Impressive work by Replit team👏
How @tibo_maker turned a $2K MRR failing product into a $600K MRR business:
"When I acquired Typeframe, it was doing $2K MRR. I spent money on this product and I just wanted to not be wrong.
This is the #1 mistake that founders make. It's more important for them to not be wrong than to be successful.
If you force the selling of your products, you're not listening to people telling you there's another opportunity that might be bigger."
After this, Tibo noticed that people wanted to make viral shorts on social media, so he pivoted the product to Revid and now it's making $600K+ MRR.
📌 Watch him talk more about it here: https://youtu.be/0UnZnonMN9o?si=42NWfMrrSIz28Da1&t=104
Peter Yang: "I shipped 9 failed products before one took off...now I'm doing $1M+/month."
Here's my new episode with @tibo_maker, a solo founder who bootstrapped 5 AI products to $1M+ / month.
Tibo walked me through his exact playbook:
✅ How to validate ideas and fail fast
✅ Why his top
Peter Yang
How @tibo_maker turned a $2K MRR failing product into a $600K MRR business:
"When I acquired Typeframe, it was doing $2K MRR. I spent money on this product and I just wanted to not be wrong.
This is the #1 mistake that founders make. It's more important for them to not be wrong than to be successful.
If you force the selling of your products, you're not listening to people telling you there's another opportunity that might be bigger."
After this, Tibo noticed that people wanted to make viral shorts on social media, so he pivoted the product to Revid and now it's making $600K+ MRR.
📌 Watch him talk more about it here: https://youtu.be/0UnZnonMN9o?si=42NWfMrrSIz28Da1&t=104
Peter Yang: "I shipped 9 failed products before one took off...now I'm doing $1M+/month."
Here's my new episode with @tibo_maker, a solo founder who bootstrapped 5 AI products to $1M+ / month.
Tibo walked me through his exact playbook:
✅ How to validate ideas and fail fast
✅ Why his top
swyx 🇸🇬: http://x.com/i/article/2047209375220076544
codex empowers anyone to build
dani: my friend was building a crazy game last night before we went out.
it’s a turn-based game that’s a mix of guitar hero, dota, and rpg mechanics.
he had sketches and a shitty godot codebase and thought getting it playable would take weeks.
i showed him Codex, we cleaned up the
codex empowers anyone to build
dani: my friend was building a crazy game last night before we went out.
it’s a turn-based game that’s a mix of guitar hero, dota, and rpg mechanics.
he had sketches and a shitty godot codebase and thought getting it playable would take weeks.
i showed him Codex, we cleaned up the
Brian Tyler Cohen
Republicans have been making this argument my entire life. We had 8 years of Clinton, 8 years of Obama, 4 years of Biden— never became a socialism. But you know what we got? More jobs, lower unemployment, higher GDP in Democratic administrations than every GOP administration in the last 3 decades. Without fail.
Laura Ingraham: At this moment, Donald Trump is the only thing standing between America and socialism. That’s why the hard Left wants him gone by any means necessary, and why they’ve bred and fed countless “recruits” to answer their twisted call.
Amit Bhatia
Agent-s does 5 out of 7 from this list pretty well. Been using it for couple of weeks and it works the best after OpenClaw stopped working for me.
I'm certain @mattshumer_ is cooking up something big!
Peter Yang: A great personal agent should:
1. Get work done across email, calendar, Google Workspace, or any API/MCP it's hooked up to
2. Act proactively and reliably (e.g., cron jobs, triggers, follow-ups)
3. Have excellent memory that helps it "just get you" over time
4. Work across
kepano
I am building Obsidian Reader because I wonder how the web would feel if it was designed solely around the reading experience.
There is so much cognitive burden that comes with every site having different layouts, fonts, ads, popups, cookie banners, engagement traps, etc.
kepano: Obsidian Web Clipper now lets you manage highlights, and stays in Reader mode when you click links.
It's such a pleasant way to browse the web. You can control the colors, fonts, and easily copy anything to Markdown.
Lulu Cheng Meservey
The shooter’s manifesto includes hypothetical objections and his rebuttals to them
This is what he thinks is one of the top objections to an assassination attempt on multiple people
Imagine seeing the world this way
New York Post: Read White House Correspondents' Dinner gunman Cole Allen's full anti-Trump manifesto https://trib.al/ytSnFnx
Gandalv
Trump has spent the better years of his second term treating allied governments like freeloading tenants who should be grateful he hasn’t changed the locks.
The result, predictably, is that they’ve stopped waiting for the lease to expire and started buying their own buildings.
What’s happening now is qualitatively different from the grumbling of 2017-2019. Back then, European leaders still believed the relationship was cyclical – that America would eventually sober up and the old architecture would reassemble itself.
That belief is gone. Greenland changed everything. For ever. What replaced it are structural decisions:
German rearmament on a scale not seen since before reunification, French-led European defence coordination that pointedly excludes Washington, South Korean quiet conversations about indigenous nuclear deterrence, Japanese defence spending hitting levels that would have been politically unthinkable five years ago.
The crucial dynamic is that these shifts generate their own momentum. A German arms industry rebuilt to supply a European force doesn’t get mothballed when the next US president calls. A generation of European defence planners trained to operate without American command integration doesn’t unlearn that. Trade relationships rerouted through new corridors don’t spontaneously return to their old paths just because someone in Washington eventually extends an olive branch.
Trump appears to have understood American power as a static asset – something you could spend down for short-term leverage without diminishing the underlying account.
What his allies understood, and what he didn’t, is that the asset was always relational. The moment they stopped believing in it, the balance sheet changed permanently.
The irony is almost classical. He wanted to make America’s allies pay more for American protection. Instead, he convinced them to stop needing it.
Fareed Zakaria: President Trump’s abuse of America's allies has reached a tipping point.
Now countries have started making long-term policy shifts — and those shifts will soon take on a life of their own.
My take:
Austen Allred
Today I learned Japan’s Kansai International Airport, which opened in 1994, has never lost a single bag.
20-30 million passengers per year for 32 years and ZERO lost bags.
Stone cold Lloyd 🫡
Lloyd Blankfein: Was at the WH Correspondents dinner last night, a rare DC trip for me without a subpoena. On the positive side—was exciting, no one was killed, and ended early. I noted a new litmus for status among the gov’t elite—whether you were whisked away by secret service, or left to fend.
Self preferencing lmao
Easton R: @garrytan Tried to do a similar thing and had Claude to help me
I know there is a note on this but I wanted to pass along this counter example
In 2014 the NYT documented remaking the Sunday sports page at 8:30 PM, 11 PM, and 12:30 AM on a Saturday night for the Belmont Stakes and Stanley Cup.
Last night at 8:36 PM someone opened fire at the White House Correspondents' Dinner. Secret Service evacuated the President. 2,600 of us were in the room.
The Sunday edition has at least two more press runs after 8:36 PM.
Neither edition ran it on A1. They ran a medical billing story instead.
Horse racing gets a remake. Gunfire in DC while all of the White House press corps is in the room… doesn't.
https://archive.nytimes.com/www.nytimes.com/times-insider/2014/06/09/in-sports-its-a-marathon-with-sprints-in-between/
Katie Miller: The third assassination attempt doesn’t make the front page of the New York Times.
Browser Use
You can now make autonomous agents in Browser Use Cloud!
Larsen Cundric: Introducing: Browser Use Box (bux). Your 24/7 personal agent box, powered by Browser Harness. ♞
We got tired of agents that vanish when you close the laptop. So we put them on a server.
> 24/7 box that runs while you sleep
> Real Chrome with persistent logins
> Telegram baked
Barack Obama
Although we don’t yet have the details about the motives behind last night's shooting at the White House Correspondents Dinner, it’s incumbent upon all us to reject the idea that violence has any place in our democracy. It’s also a sobering reminder of the courage and sacrifice that U.S. Secret Service Agents show every day. I’m grateful to them – and thankful that the agent who was shot is going to be okay.
Mada Seghete
It doesn’t get better than @swyx singing @Harry_Styles @projecthailmary Sign of the times in his @baseten shirt! Watching on repeat 😝🫠😘😬
Vox
gstack 1.15 just dropped.
lately i'm in two gstack commands every day: /design-consultation and /design-review. with codex and claude code in the loop, the design step is way more efficient. this release garry also put the whole stack on a diet:
→ per-invocation tokens down 25%, the heavy plan-ceo preamble cut 43% (54KB → 31KB)
→ new 654-line real-terminal harness for testing AI back-and-forth
→ codebase shrank by 11.6k lines
→ added 34 new tests (11 E2E + 23 unit)
gstack runs your dev workflow, gbrain feeds your agents the code.
Garry Tan: GStack v1.15.0 just hit main
I heard you, GStack got a little fat so it lost 25% to 43% of its token usage. Sometimes you add features. Sometimes you compress the tokens. I had to build a real TTY test harness using Bun to test AskUserQuestion but it worked!
Coding agents will be the foundation of all superintelligence.
At a minimum, coding ability is indistinguishable from 'proficiency with computers'. Great coding agents like Claude Code master bash, filesystems, configuring and installing programs…
But it's also about self-improvement. A coding agent has the ability to examine its source, its state, its skills, its instructions… it can propose changes to itself (with human supervision and audit trail, I recommend), or even mutate itself directly.
In retrospect, this should be obvious. "What I cannot create, I cannot understand". Coding fluency has given models a deeper understanding of all computer and knowledge work. To master programs, you must be able to create them.
Lee Robinson: It wasn’t obvious to me one year ago that an excellent coding agent would also be the path to a general agent for all knowledge work.
But now it makes a lot of sense. I’m interested to see where AI is at next year and what seems obvious then in retrospect.
Aghion Philippe
👉 Sur l’#IA, mon optimisme est prudent. Je reconnais les dangers de cette technologie, mais si on en freinait le développement en Europe, au nom du principe de précaution, son essor aurait lieu ailleurs. Il faut l'exploiter au mieux.
👇 Discussion dans Le Monde
Our Principles:
Democratization, Empowerment, Universal Prosperity, Resilience, and Adaptability
https://openai.com/index/our-principles/
x.com