Great, I can now combine the potential maliciousness of a script with the potential vulnerabilities of an AI Agent!
Jokes aside, this seems like a really wierd thing to leave to agents; I'm sure its definitely useful but how exactly is this more secure, a bad actor could just prompt inject claude (an issue I'm not sure can ever be fixed with our current model of LLMs).
And surely this is significantly slower than a script, claude can take 10-20 seconds to check the node version; if not longer with human approval for each command, a script could do that in miliseconds.
Sure it could help it work on more environments, but stuff is pretty well standardised and we have containers.
I think this part in the FAQ wraps it up neatly:
"""
What about security? Isn't this just curl | bash with extra steps?
This is a fair concern. A few things make install.md different:
Human-readable by design. Users can review the instructions before execution. Unlike obfuscated scripts, the intent is clear.
Step-by-step approval. LLMs in agentic contexts can be configured to request approval before running commands. Users see each action and can reject it.
No hidden behavior. install.md describes outcomes in natural language. Malicious intent is harder to hide than in a shell script.
Install.md doesn't eliminate trust requirements. Users should only use install.md files from sources they trust—same as any installation method.
"""
So it is just curl with extra steps; scripts aren't obfuscated, you can read them; if they are obfuscated then they aren't going to use a Install.md and you (the user) should really think thrice before installing.
Step by step approval also sorta betrays the inital bit about leaving installing stuff to ai and wasting time reading instructions.
Malicious intent is harder to hide, but really if you have any doubt in your mind about an authors potential malefeasance you shouldn't be running it, wrapping claude around this doesn't make it any safer really when possible exploits and malware are likely baked into the software you are trying to install, not the install.
tldr; why not just have @grok is this script safe?
This is some really fantastic feedback, thank you!
I personally think that prose is significantly easier to read than complex bash and there are at least some benefits to it. They may not outweigh the cons, but it's interesting to at least consider.
That said, this is a proposal and something we plan to iterate on. Generating install.sh scripts instead of markdown is something we're at least thinking about.
I think relying on the vocabulary to indicate AI is pointless (unless they're actually using words that AI made up). There's a reason they use words such as those you've pointed out: because they're words, and their training material (a.k.a. output by humans) use them.
No American used "delve" before ChatGPT 3.5, and nobody outside fanfiction uses the metaphors it does (which are always about "secrets" "quiet" "humming" "whispers" etc). It's really very noticeable.
The link you posted doesn't back up the statement that "No American used "delve" before ChatGPT 3.5". Instead it states that _few_ people used it in _biomedical papers_. I've seen it (and metaphors using the other words you noted) used in fiction for my entire life, and I sure as hell predate chatgpt. This is why it's a bad idea to consider every use of particular words to be AI generated. There are always some people who have larger vocabularies than others and use more words, including words some people have deemed giveaways of AI use.
That said, their use may raise suspicion of AI, but they are _not_ proof of AI. I don't want to live in a world where people with large vocabularies are not taken seriously. Such an anti-intellectual stance is extremely dangerous.
I've been reading deep research results every day for months now and I promise I know what AI writing style looks like.
It has nothing to do with "large vocabularies". I know who the people with large vocabularies were that originally caused the delving thing too, and they weren't American. (Mostly they were Nigerian.) I'm confused what you think specific kinds of metaphors involving sounds have to do with large vocabularies though.
> I've seen it (and metaphors using the other words you noted) used in fiction for my entire life
And the point is that this article isn't fiction. Or not supposed to be anyway.
People with large vocabularies tend to be heavy readers, and therefore experiencing these words and metaphors more than people with smaller vocabularies. I think there's a direct link between people attempting to use certain words as proof of AI and the fact the younger generations aren't reading as much as older generations.
Somewhat contradictory, I don't think you can ignore fiction when discussing technical writing, since technical writing (especially online) has become far more casual (and influenced by conversation, pop culture, and yes, even fiction) than it ever was before. So while as I noted above, younger people are reading less these days, people are also less strict about how formal technical writing needs to be, so they may very well include words and expressions not commonly seen in that style of writing in the past.
I'm not arguing that these things can't be indicators of AI generation. I'm just arguing that they can't be proof of AI generation. And that argument only gets stronger as time goes on an more people are (sadly) influenced by things AI have generated.
But now Americans do use "delve" since 3.5. So what? No Americans used "cromulent" as a word either until Simpsons invented it. Is it not a real word? Does using it mean the Simpsons wrote it?
I am not sure about the use case but I was happy to see they advertise a long battery life. Still, as soon as I learnt that it is disposable I lost any interest in the product.
I very likely wouldn't have bought it anyway but I am surely not going to buy disposable tech.
How much rechargeability really requires from hardware point of view?
I suppose the problem is that there are no standard for tiny magnetic chargers/cables. Every watch comes with their own, and they need be custom designed. For a device this large as much of the charging electronics should be outside the ring.
And another (small?) problem is that you'd need to electrically protect those external pins.
I'm with you, a little pogo-pin connector with all the charging circuitry in the external dongle would add, I don't know, ten dollars to the BOM. Very cool product but I won't buy a gadget I know I'd dispose in two years.
Yeah, this would have been a neat tool for "middle-of-the-night" thoughts, but I don't want a disposable electronic device. I get why it is that way; not having any charging hardware probably makes the device much smaller. But I'd rather it be a bit bigger and rechargeable.
"Years of average use" is great until you realize that it actually means "Roughly 12 to 15 hours of recording".
Not sure how long my iphone can record for, but it's probably close to that. Afterwards I get to charge my phone instead of recycle it, though.
Apple, don't hire this man.
Edit for the downvoters: can my iphone not record that long or something? iphones can't recharge? Just hate Apple and love e-waste rings? Enlighten me.
At the lower 12-hour end, if you're doing ~10 seconds per recording (remember this device is primarily for very short reminders and quick commands), that's ~4.3k recordings total. Also keep in mind that you'd only use it when it's inconvenient/undesirable to reach for your phone or any other device, so it's possible this may only be used say 5x in a day at most on average (likely far less). Which means ~2.5 years worth of usage at the lower end, and you'd only ever have to take it off if going for a swim.
Contrast to a phone that, though it has far more capability, you'd have to remember where it is before even reaching for it wherever, and usually has to be on a charger for anywhere from 30 minutes (with super charging) to a few hours daily. Or even being at a laptop/desktop, and at least having to open the relevant app, type/talk into it and then close again to return to primary task. The ring is an instant win for 24/7/365 convenient presence.
You have to trust it will actually get recycled though. I struggle to believe they'll be swapping out the batteries and reselling these as reconditioned. (I struggle to believe many people will even send them in for recycling tbh.)
The environmental benefit of sending it in for recycling is probably negated by transporting it all the way back to them for starters. Better to just drop it at the local ewaste collection facility. They'll be less specialised but there isn't a lot of material in it.
I guess there's a market for it and in the scale of things it isn't so bad: you could make 10 disposable vape sticks from the materials in one of these rings. And they're expensive enough that they'll never sell more than 100k or so of them. Relatively speaking it's no measurable impact.
For me it's more a matter of principle though. As a society we frown on disposable gizmos these days and for good reason.
This is the man responsible for the first death of Pebble. His product vision is not good, and he is heading straight for another wall with his current mistakes.
I don't really know what it's like in America but 150k here is insane.
If you really can't land a job, it's either time for an attitude adjustment or perhaps a third bite at the apple.
I don’t know where you are but 150k is decent at best in most US and Western European big cities. So if someone has a few years of experience under their belt, the expectation isn’t too complacent.
US$150k is not only "decent" in high-income Western European cities (Paris, London, Amsterdam, etc.), it's a huge salary, usually in the top 5% of earners.
This expectation is why software development has become the neo-yuppies of the 2010s-2020s, people with overinflated expectations because they are keeping-up-with-the-Joneses from the internet being loud about their high salaries. Yes, there are people earning a lot in this industry but it's definitely not the norm outside of the inflated US salaries, the USA is the outlier.
I literally don’t understand the rationale behind trying to bring down anyone aiming for a higher wage.
I live in Europe, and 150k isn’t a moonshot salary here. Just because something isn’t the norm doesn’t mean no one should try to achieve it.
Also, OP doesn’t live in Europe and clearly doesn’t care. So 200k is a reasonable ask there. American salaries aren’t outliers; they correspond to the cost of living. Being bitter about your own wage and trying to denigrate someone else isn’t helpful.
> Also, OP doesn’t live in Europe and clearly doesn’t care.
My point was to what I stated about European cities, not sure why you felt the need to state the obvious, refrain from strawman-ing, please.
> I literally don’t understand the rationale behind trying to bring down anyone aiming for a higher wage.
One can aim as much as they want but there's reality: that kind of wage is reserved for less than 5% of the population of already very rich countries, by definition it's not achievable by the majority, and having that as an expectation is bound to bring a lot of people feeling like failures because they can't achieve it. Keeping reality in mind is very good to not end up frustrated because you are earning a salary of 80k€ which is already more than 70-80%+ of what workers earn in the EU...
Aim for what you feel like while keeping in mind that, statistically speaking, you might not achieve it but can still have a very good life with less than that.
I don't really have any interest arguing what a respectable salary on HN is. The recruiter I was working with clearly stated that $180-200k given my resume was reasonable. America isn't europe.
I sold my house last year. I looked up the average price growth over the 8 years I owned it for, did some maths and came up with a value. Estate agents even vaguely corroborated it - they thought it was high but within range.
Then I started having viewings on the house, and no one liked it at the price. People had really weird comments: at this price the garden should be bigger, there aren't enough rooms downstairs, the staircase is too wide. Objectively, whatever that means, it was a lovely house - good location, modern, good condition, nice garden, light and spacious. Nothing wrong with it.
Long story short, in the end I sold it for 15% less than the "fair market price" I came up with, and also 10% below the real estate agents estimate.
It stung, but it's a reminder that the market value is what someone is willing to pay. If no one wants to pay my fair market value then it's not a fair market value, period.
I'm just a rando on the internet, but I think the attitude that you are definitely "worth X" but no one wants to pay you this might be counterproductive. Your empirical evidence says you are apparently worth Y. Markets change, recruiters lie etc.
Start with that as as an evidential baseline and ask yourself what could increase your "value" - be it how you present your experience, how you interview, or what skills you could add or niche you could explore.
I'm gonna be real with you: you need to face the fact that you don't understand the game you're playing.
I know because I was there. I founded a startup that I can make sound impressive. I left when I burnt out, thinking of course what I did was impressive enough to easily land a job. I floundered for 5 years.
Places look at your resume and either see someone who is overqualified (they're right) or someone who hasn't worked on a real project (they're right). If you can't swallow that, you're going to keep beating your head against the wall until you do.
I started refusing any interview that had a live coding exam (I vibed way better with places that did project based exams). I coded my own projects to hone my skills.
Eventually I decided I needed to play the game. What I did was really niche. I took a web dev boot camp and practiced code exams.
Now I have a job where I started at a low salary, but they created a senior position for me in less than a year. It's looking like managing my team will be a thing before year 2. I will get paid less than other people in the industry, but if that ever really bothers me I'll be able to move laterally and make more easily.
This is all to say: suck it the eff up. Your experience is useful, but you're playing a different game. Lower your expectations, get the chip off your shoulder, and maneuver whatever you can find into a win.
Jokes aside, this seems like a really wierd thing to leave to agents; I'm sure its definitely useful but how exactly is this more secure, a bad actor could just prompt inject claude (an issue I'm not sure can ever be fixed with our current model of LLMs).
And surely this is significantly slower than a script, claude can take 10-20 seconds to check the node version; if not longer with human approval for each command, a script could do that in miliseconds.
Sure it could help it work on more environments, but stuff is pretty well standardised and we have containers.
I think this part in the FAQ wraps it up neatly:
""" What about security? Isn't this just curl | bash with extra steps? This is a fair concern. A few things make install.md different:
Install.md doesn't eliminate trust requirements. Users should only use install.md files from sources they trust—same as any installation method. """So it is just curl with extra steps; scripts aren't obfuscated, you can read them; if they are obfuscated then they aren't going to use a Install.md and you (the user) should really think thrice before installing.
Step by step approval also sorta betrays the inital bit about leaving installing stuff to ai and wasting time reading instructions.
Malicious intent is harder to hide, but really if you have any doubt in your mind about an authors potential malefeasance you shouldn't be running it, wrapping claude around this doesn't make it any safer really when possible exploits and malware are likely baked into the software you are trying to install, not the install.
tldr; why not just have @grok is this script safe?
Ten more glorious years to installer.sh
reply