A critical LangChain core bug can turn a single LLM prompt into a secret-exfiltration or object-instantiation exploit. If you run LangChain in production, this is urgent.
I have seen many LLM devs' encountered this at some point. Good to see that you are not only pointing out the inconsistency but also actively advocating a common benchmark.
It's hard to believe that the OneDrive File Picker still doesn't have fine grained OAuth scopes in 2025. Allowing read access to the whole drive just to upload one file goes against the principle of least privilege.
> It's hard to believe that the OneDrive File Picker still doesn't have fine grained OAuth scopes in 2025
We are talking about Microsoft here.
Me gets a new laptop, company allows SW instalation only from an internal portal, i don't find Teams there, i ask a colleague how does one one install Teams: IT said to download it from microsoft.com. ROTFL. Of course it does not need "elevated priviledges" to install. Of course it is installed for every account on the computer and has access to all user files.
But some people still _believe_ the (first appeared in Win95) "most secure Windows ever" lie.
Sad to see Pocket go. It's been part of my daily routine. That said, this list of alternatives is gold. I'm leaning toward Matter or Readwise Reader right now, but curious what others are switching to. Anyone found a hidden gem worth trying?
I just got used to Arc and now it’s getting sunsetted already? I get the learning curve bit, but Arc really felt like a fresh take on browsing. I hope Dia lives up to the hype — anyone here tried the alpha yet? Would love to hear your impressions!
Zen browser is very promising as a hopefully long-lived alternative to Arc with essentially the same design. It’s missing a few important features (like folders, which should be coming any week now), but overall it performs more smoothly than Arc on windows. Though I’m still using Arc for work on my Mac (for now).
Dia is an “AI native browser,” and the Arc CEO said that browsers are going to be killed by AI chat. So if you want to get real work done, Dia is probably not going to be an option.
Seconding that Zen has been an amazing Arc replacement for those happy with Firefox based browser.
The only challenge is daily driving both Zen and Zed as a code editor is a pain when it comes to macOS spotlight (/raycast).
might not meet your needs but I had similar issues and found an app called rcmd, it lets me easily bind keybinds to my commonly used apps, and you can also easily toggle between multiple windows of the same app via the keyboard
They’re not sunsetting it; they’re just keeping it in the same maintenance mode they told us about a year ago when they said the 2.0 browser would be getting all the new development effort. If you like how Arc works right now, you’re good.
I don’t know why it’s being reported like a new situation or a shutdown. You can see in the article comments that the author gets called out for misunderstanding and doubles down.
The majority of implementations use Python or Node.js directly, which keeps things quick and simple. I like how it is extremely easy to extend using your own tools and uses JSON-RPC instead of standard interface.
Based on your recent comment history vibes, I'm pretty sure that you're using a LLM to post comments; and the remainder is self-promotion towards your own articles. Please don't do that here; HN aims to have interesting and not generic conversations: https://news.ycombinator.com/newsguidelines.html
fwiw I don't have a problem with LLM posts. But I do agree that this is pretty generic. If you want to use an LLM to post comments: find a better prompt / workflow
LLM-generated comments are not wanted on HN; we want to preserve it as a place for discussion between humans.
Also, comments of the format "here's what an LLM said about this topic:" are best avoided. We don't want to normalise a style of discussing issues in which we generate an LLM output and make that the central conversation topic; we prefer original human thought here.
I guess if we're gonna do monkey's paw/work to rule type of interpretations, HN should just add "we value authentic human experiences as opposed to posting simply for the sake of it" in the guidelines. I'll shoot off an email to the mods about this later today.
I’d much rather a useful AI-aided post that gives me insight than the almost daily pedantic unrelated gripe “interesting post but I hate the font”. This AI slop is bad but is it worse? In any case “bad” should be a reasonable bar to get over.
