Top 30 stories from Hacker News. Top 10 include comment highlights. Compiled at 20:11 UTC.
364 points by jhoho · 145 comments
What xAI Grok Build CLI actually sends to xAI - a wire-level analysis (grok 0.2.93) - grok-build-cli-wire-analysis.md
What HN said:
phaseleza: I always separate the coding tools from LLM providers, and use bubblewrap to sandbox the coding tools so they: 1. Can only read the working project directory, with .git read-only and sensitive directories hidden (mounted as empty directories). 2.
gitgud: This is one of the reasons why native proprietary coding agent runners like claude-code, codex, grok-build etc are so dangerous for privacy… you just don’t know what “secret sauce” they’ll add in the next update… It’s much safer to use something like opencode and use models via t...
freakynit: "It uploads the whole repository — every tracked file's content plus git history — independent of what the agent reads" Holy cow!!!! I mean I kinda expected Elon would do something like this to try to catch-up.. but this is extremely concerning.
Karmakosmik: Isn't that expected? I always assumed the agent owns (at least) the current workspace (whatever dir it's launched in) and so can do whatever it wants in there. If they actually use this try and do things in the backend and saving prompt RTTs and tool calls that would be in my int...
361 points by subset · 103 comments
What HN said:
recursivedoubts: Building visualizations with LLMs has been a major boost for my CS classes: https://htmx.org/essays/universities-and-ai/#demos-visualiza... Many visualizations that I have always wanted but just didn't have the time to build, I now have.
luciana1u: Terry Tao using coding agents to build apps means we're one step away from a Fields Medalist asking an LLM why his Docker container won't start, just like the rest of us.
semiquaver: There is infinite latent demand for software, most especially outside the traditionally software-focused spaces. If LLMs stopped improving today it would take us 10 years to catch up to the new software-writing abilities that have become available.
wffurr: Nice balanced perspective there at the end: "as such [LLM-coded interactive] supplements are not mission-critical to the core of the paper, I again feel that the downside risk of using guided interaction with LLM agents to generate such visualizations is acceptable." It's a tool.
259 points by compiler-guy · 148 comments
Vinton Cerf, one of the creators of the protocols underlying the internet, will step down as Google's chief internet evangelist next week.
What HN said:
jvanderbot: I met Vint at the Kech Institute for Space Studies. He arrived to help us look at in-space data centers for planetary science throughout the solar system. He was a big proponent of delay-tolerant networking and other useful networking stacks, so he was the "rep" for that layer of...
mehulashah: For all the naysayers in this thread, I gotta say you’re wrong. Vint is a class act. Humble, helpful, and optimistic. Not to mention one of the most impactful computer scientists of our generation.
Angostura: I interviewed him a few times, when I was a tech journalist in the 90s - a very impressive man. However I never forget my surprise, Idly flicking through TV one evening and coming across Earth Final Conflict - and there was Vint in a fairly substantial role
djtriptych: hah. I was an intern at Google in 2005 when he was hired and remember the wave of reverence that went through Mountain View. Salute to a legend! It’s like two lifetimes in tech years.
241 points by signa11 · 40 comments
What HN said:
dakra: Hi! Maintainer of Ghostel here. baokaola and I actually wanted to do a "Show HN" next week, but looks like someone was faster submitting the link. Have a look at the GitHub repo which is a bit nicer for a quick overview: https://github.
LtdJorge: I think the title should mention Emacs somewhere. A terminal emulator is different than a terminal emulator for Emacs.
jdormit: I recently switched from vterm to ghostel, and it is generally much, much better - noticeably faster (e.g. fancy TUI apps that try to refresh the whole terminal every frame actually work), more reliable input handling, and a nicer ELisp API.
varjag: Been using it for bit over a month now. It's really nice that you can click on code references in Codex summaries and open them right there in Emacs buffers.
194 points by systima · 104 comments
Claude Code vs OpenCode token overhead measured at the API boundary. Out-of-the-box baselines, instruction file weight, MCP schema tax, subagent multipliers, and cache-write behaviour.
What HN said:
mcv: What really burns tokens is sub agents. I once gave Claude Code a pretty big task, and it immediately launched 7 sub agents which burned through my budget before even one of them was finished. Tried again 5 hours later: same result.
korrectional: My opinion is that claude code uses more tokens simply because Anthropic makes more money that way and forces people into their subscriptions. This is supported by the fact that they won't let you use your sub on a different coding agent. I use pi btw.
systima: UPDATE: After reading PUSH_AX's valid comment:
This is like saying contractor (A) asked for $33,000 to undertake the work and contractor (B) asked for $7,000 Are we measuring and caring about the right thing?We will update the post to include: 1) A more in-depth task.
estetlinus: Recently switched to Codex after 6m in Claude. Codex seems more open, it’s easier to follow what the model is doing and the approvals have a better UX. Overall, it just feels more transparent. Cost of switching was close to 0.
188 points by silcoon · 104 comments
What HN said:
aaronbrethorst: I have an almost-four year old child and not a lot of downtime. I used to listen to podcasts when I was doing dishes, cleaning the house, walking the dog, etc. I've mostly abandoned podcasts in favor of audiobooks.
goodroot: Love this blog, appreciate the author. > This is probably the most difficult part. I had to remove all social media and streaming apps from my iPhone. I removed Instagram, YouTube, Facebook, etc.
5555624: Get a long book, a timeless classic, and read one chapter a night. I grew up reading all the time. About 20 years ago, I found myself reading less and less. I decided to read "The Count of Monte Cristo" again.
asimpletune: I agree with everything this guy said and yet I still have a totally different take. Some of my friends and I started a book dedicated to reading books in the public domain (https://b00k.club).
150 points by saikatsg · 110 comments
What HN said:
robot_jesus: I'm in my 40s with genetic predisposition for Alzheimer's. Been seriously considering the past year or two paying out of pocket for Shingrix. I think it would be ~$500 total for two doses.
hereme888: Replicated association, which is strong, but not proof. Initial study saw a 3.5% absolute reduction in dementia diagnoses over seven years with a very wide confidence interval. In Australia the study was replicated with 1.8% absolute reduction over 7.4 yrs.
modeless: Unfortunately this is a spurious finding. See this presentation: https://youtu.be/qlTnnQytOJ0 The mechanism is that people with the shingles vaccine are less likely to visit the hospital (because they don't get shingles).
what_hn: Is it also possible they're finding healthier people that are proactive in their treatment, maybe even exercise more, work longer, etc...
142 points by zdw · 70 comments
What HN said:
ChiperSoft: This analogy has layers that OP probably doesn't recognize. The film industry loves cgi because none of the digital vfx houses are unionized and they can treat the artists like crap. It severely devalued a ton of skilled labor around miniature and set design.
singpolyma3: I mostly like this article but > Those who refuse to use an LLM will fall behind because they won't be able to produce as much Seems like a silly and needlessly aggressive take. Fall behind what? Able to produce "as much" what? I've never been evaluated on volume in my life.
lnrd: > Writing every line by hand is no longer the norm. Those who refuse to use an LLM will fall behind because they won't be able to produce as much > It remains important to be able to read the code and understand the architecture.
konart: >In 1993, Jurassic Park came out and revolutionized the use of CGI in films I thought Terminator 2 did this. At least around where I lived everyone talked about T2 and questioned how they did this or that. JP? Not really.
128 points by AlexeyBrin · 67 comments
Learn the Odin Programming Language and demystify low-level programming
What HN said:
pseudony: Having fun with this. Never bought into rust (have studied, have a (mostly AI-generated app in rust). Wrote some Zig but Odin is even less overhead for me. I first loved Zigs built-in build system but having tried to wrap/use C libraries from both, I must say I prefer Odin.
andyfilms1: I've been using Odin for about 6 months now, and to be honest, it's hard to find fault with it. I've used it for STM32 microcontroller firmware, web and desktop applications, and all are performant and compile quickly.
gcanyon: I wonder if this will help them get a wikipedia page. (not sure whether this comment is a joke or serious...)
ant6n: Kinda wanna know why I should learn this language. Unfortunately there’s no wiki entry (deleted with some controversy abut notability), so it’s hard to get the gist of it.
123 points by zaikunzhang · 90 comments
What HN said:
overgard: I think we're finding measures of "productivity" in almost all fields are pretty bad and AI is a great way to game them. PR's, papers, etc. We have to stop looking at volume-of-stuff as a useful metric.
dahart: > Scientists who adopt AI gain productivity and visibility: On average, they publish three times as many papers, receive nearly five times as many citations, and become team leaders a year or two earlier than those who do not.
Labo333: > “It’s not about the architecture per se,” Evans says. “It’s about the incentives.” It would have been useful to check whether less original work was already getting more citations before AI adoption.
skeledrew: As with other fields touched, AI is merely amplifying what was already there. The aim of many scientists isn't discovery in and of itself. Discovery is a side effect of their primary drive to publish and - hopefully - become well known.
AI/ML
Other
Programming
Science
Stories and comments sourced from Hacker News public API. Not affiliated with Y Combinator or Hacker News.