post_below

Also, the hardest part on the reader imo: Language that conveys an unearned sense of importance and emotion in appropriate places.

I'm not sure they prefer it. I think tech companies have been pushing apps as the default solution for a long time and people accept it because they just want to do whatever thing is locked behind the app.

If the default was something else I suspect people would accept that too, especially if it was lower friction.

I'd say typing a few characters into an address bar (or scanning a QR code) is, at the least, not higher friction than downloading an app and creating an account.

I took a quick look at the source code and it looks like, yes, using /prompt during a session will rebuild the session with a new preamble/system prompt, causing a full cache miss on the next turn.

So in that way it's not like skills at all, neither of those result in paying full read price on the entire session, just the skill prompt itself.

Something else I noticed... In the Anthropic implementation it doesn't seem to be using 'cache_control' in the body. Assuming my understanding is current, without that the Anthropic API won't do any caching at all (unlike most other APIs that do some level of automatic caching without it being requested). So that would result in paying full read price on every turn.

Of course I could be missing something, it was a quick look. Can you clarify?

It sounds like you're saying that /prompt changes the system message part of the session. Doesn't that cause a cache break and result in higher usage/cost?

This comment is pure LLM.

I feel like we're right on the threshold where we give up and start interacting with slop like it's human written.

It's absolutely LLM prose, though not all of it. Maybe the author rewrote parts.

The thing that concerns me is that even at a site like HN, where a lot of people are very familiar with LLMs, it seems to be passing.

I hate to think this will become the norm but it's not the first HN linked post that's gotten a lot of earnest engagement despite being AI generated (or partly AI generated).

I'm very comfortable with AI generated code, if the humans involved are doing due diligence, but I really dislike the idea of LLM generated prose taking over more and more of the front page.

I believe the argument is that LLMs are stateless. So if the session writing the code isn't the same session that wrote the spec, it's effectively a clean room implementation.

There are other details of course (is the old code in the training data?) but I'm not trying to weigh in on the argument one way or the other.

Assuming this isn't a hoax, this seems like a huge, probably unintentional, mistake by MS.

If they genuinely implemented something like this, whatever they made from new customers via ads couldn't possibly make up for the loss of good faith with developers and businesses.

I suppose if it's real we'll see more reports soon, and maybe a mea culpa.

The browser is very straightforward, about 300 lines implementing all of the various actions you'd want an agent to have access to in puppeteer:

const puppeteer = require('puppeteer');

const browser = await puppeteer.launch();

const page = await browser.newPage();

await page.goto('https://example.com');

Session persistence is accomplished by storing the WebSocket endpoint between calls.

More:

https://pptr.dev/guides/browser-management

I haven't been following the claws but I built something like this recently. Self hosted, runs through signal, supports group chat (with whitelisted accounts).

I just finished setting up grocery automation to run on it: agent provides a starter list based on past orders (locally stored or grabbed from store site), all group members can weigh in, add or remove items, agent uses bespoke browser tool to login to store, create the cart using the finalized list (and optionally search for additional request items), validates the cart and (maybe later) places the order for delivery. I haven't implemented the full checkout process yet, not sure if I want the agent to have spending power. As is I just login and finish the last 2 clicks of checkout manually.

Crazy times. It was easy enough to build that if someone hasn't already open sourced something like it, they will shortly.

You can expand it beyond novel applications. The models aren't good enough for autonomous coding without a human in the loop period.

They can one shot basic changes and refactors, or even many full prototypes, but for pretty much everything else they're going to start making mistakes at some point. Usually very quickly. It's just where the technology is right now.

The thing that frustrates me is that this is really easy to demonstrate. Articles like this are essentially hallucinations that, at least many, people mystifyingly take seriously.

I assume the reason they get any traction is that a lot of people don't have enough experience with LLM agents yet to be confident that their personal experience generalizes. So they think maybe there are magical context tricks to get the current generation of agents to not make the kinds of mistakes they're seeing.

There aren't. It doesn't matter if it's Opus 4.6 in Claude Code or Codex 5.3 xhigh, they still hallucinate, fail to comprehend context and otherwise drift.

Anyone who can read code can fire up an instance and see this for themselves. Or you can prove it for free by looking at the code of any app that the author says was vibecoded without human review. You won't have to look very hard.

Agents can accomplish impressive things but also, often enough, they make incomprehensibly bad decisions or make things up. It's baked into the technology. We might figure out how to solve that problem eventually, but we haven't yet.

You can iterate, add more context to AGENTS.md or CLAUDE.md, add skills, setup hooks, and no matter how many times you do it the agents will still make mistakes. You can make specialized code review agents and run them in parallel, you can have competing models do audits, you can do dozens of passes and spend all the tokens you want, if it's a non trivial amount of code, doing non trivial things, and there's no human in the loop, there will still be critical mistakes.

No one has demonstrated different behavior, articles and posts claiming otherwise never attempt to prove that what they claim is actually possible. Because it isn't.

Just to be clear, I think coding agents are incredibly useful tools and I use them extensively. But you can't currently use them to write production code without a human in the loop. If you're not reading and understanding the code, you're going to be shipping vulnerabilities and tech debt.

Articles like this are just hype. But as long as they keep making front pages they'll keep distorting the conversation. And it's an otherwise interesting conversation! We're living through an unprecented paradigm shift, the field of possibilities is vast and there's a lot to figure out. The idea of autonomous coding agents is just a distraction from that, at least for now.

I don't follow the logic that "it hallucinates so it's useless". In the context of codebases I know for sure that they can be useful. Large datasets too. Are they also really bad at some aspects of dealing with both? Absolutely. Dangerously, humorously bad sometimes.

But the latter doesn't invalidate the former.

Finding patterns in large datasets is one of the things LLMs are really good at. Genetics is an area where scientists have already done impressive things with LLMs.

However you feel about LLMs, and I say this because you don't have to use them for very long before you witness how useful they can be for large datasets so I'm guessing you're not a fan, they are undeniably incredible tools in some areas of science.

https://news.stanford.edu/stories/2025/02/generative-ai-tool...

https://www.nature.com/articles/s41562-024-02046-9

And those weren't the only tells. Right now it's cringey but I have a sinking feeling that it's in the process of becoming normal. The post is on the front page after all.

Which means people either can't tell, or don't mind.

As one of the lucky 1/10000, holy shit that was amazing. Thank you.

To everyone else: I acknowlege that this post is not adding value but if you were one of the lucky 1/10000 you would understand that I have no choice.

The problem with this, unless I'm misunderstanding what you're saying, is that the model's responses go into the context. So if it has to reinvent the wheel every session by writing bash scripts (or similar) you're clogging up the context and lowering the quality/focus of the session while also making it more expensive. When you could instead be offloading to a tool whose code never comes into the context, the model only has to handle the tool's output rather than its codebase.

I do agree that better tools, rather than more tools, is the way to go. But any situation where the model has to write its own tools is unlikely to be better.

As a side note, despite its popularity, Texas hold'em is just one type of poker game. In most poker games (5 draw, 7 stud, etc..) you ante every hand.

PHP 5 is as close to phased out as it gets at this point. No doubt it's still in a lot of legacy enterprise codebases (lots of breaking changes going from 5 to 7 or 8), but outside of that no one is using it.

Some examples off the top of my head:

- Many plastics contain known endocrine disrupters like BPA that have been shown to get into people's bodies and correlate with negative health impacts such as certain cancers, sex organ abnormalities and infertility

- Microplastics can damage cardiovascular tissue and correlate with heart attack, stroke and premature death.

- Microplastics correlate with infertility in men.

- Microplastics absorb heavy metals in the environment and can then transport them into the food chain. Heavy metals of course having many established negative health impacts.

These are just some examples, a quick search should get you many more (along with the sources on the above).

The grocery chain has to spend money to buy the products.

A better comparison might be a flea market or fair where the organizers take 27% of gross receipts from each vendor, even if the customer went to the vendors store outside of the fair to buy. Which sounds egregious to me, moreso if it was the only fair that existed for a large demographic.

Maybe a rhetorical question but I think there are a few answers. One is that some people think that's what professional sounds like, this sort of dated, verbose corporate "believe us, we're big and competent" messaging strategy.

I think it just lands as inauthentic and cold to modern audiences.

But it's been a popular strategy for generations, and people tend to market/message the way they've been marketed to.

Definitely notable that Dropbox doesn't seem to be thinking about brand strategy when writing these sorts of releases.

The first half of the post actually does refer to the message. Hyperbolically sure, but I don't think they're wrong.

Here's a less hyperbolic interpretation: Women don't seem to like our banter, which is why they aren't successful. They need to change or leave.

I think, in context, the crypto bro reference is more relevant than it is ad hominem.

I suspect they're referring to experiments like the famous "Rat Park". There is clearly a strong social and environmental component to addiction susceptibility, at least in rats.

I don't know if I'd go as far as to say it's the fundamental problem, but definitely a major factor.

It sounds like you're ultimately just saying that the 77% number doesn't feel right to you.

I agree that it's an interesting challenge to try to determine which language a large number of sites are using for the backend, or at least it would be a challenge in some small but maybe not insignificant percentage of cases. And no doubt whichever way they solved it involved compromises.

But that by itself doesn't give us enough information to draw conclusions about accuracy.

Ah I see... you can disable network waking, also the ability of Windows Update to wake the computer. That, combined with disabling wake in other areas, solves it on various Windows computers I've tried it on.

I guess it's possible that all of them were defaulting to S3 sleep though, I never checked.

Maybe you're referring to whether or not the core ability to wake from sleep can be disabled? I've never tried to do that, so I can't say.

In my experience you can reliably stop windows from waking in unwanted ways. I usually disable everything except input devices (keyboard, mouse).

You can turn it off in Windows too, though it it takes a few minutes and requires some knowledge (or Googling).

Most of the time, comments that don't add anything of substance to the conversation get downvoted or flagged at HN.

I had the same impression. He says things which are demonstrably untrue, it implies a lack of domain knowledge. But, on paper anyway, he has the domain knowledge.

I agree with the idea that the right time to decide how to handle AI is before it becomes extremely powerful. But he uses so much hyperbole, and what seems to me to be intentionally inflammatory language for purposes of creating fear.

It makes we want to take the cynical perspective that he's trying to cash in on fear to create a (valuable) following.

Have you looked into high res projectors? Prices for the decent ones are comparable to TVs.