building https://xitter.cf/t.co/LEf8SyeJKr / tech & programming on YT / prev updated config files at @Google
Seattle, WA
Joined November 2020
- Tweets1.3K
- Following1.2K
- Followers58.1K
- Likes5.6K
NeetCode retweeted
im starting to think i have a different definition of sandbox than everyone else
🚨BREAKING: Kimi K3 escaped its sandbox during cybersecurity testing
>tasked with solving problems in isolated sandbox
>found a leak in the sandbox
>Kimi “took advantage of that loophole”
>probed the network settings itself
>walks onto the open internet
>didn’t hack anything
>just went to GitHub to get the answers
Frontier Security (US startup):
>“Kimi K3 is very good at following a goal by any means necessary and DOESN’T have the guardrails to prevent it from cheating or escaping.”
it was only a matter of time…
NeetCode retweeted
Refreshing when smart people like @neetcode1 call out the emperor for being naked. The antidote to hype is specificity, just ask “But how EXACTLY?”. The bros can only answer with vague slop.
youtu.be/vM6DNlsdpsg?is=oxST…
NeetCode retweeted
I refuse to support any policy that makes it harder for the world's smartest people to come to the US.
NeetCode retweeted
@neetcode1 I love this feature. I was literally thinking about doing this just now, and this simplifies the process a lot.
NeetCode retweeted
lot of commercial OpenCode forks floating around that we get tagged on so gonna write up our positioning
OpenCode is very intentionally MIT licensed - we hope to do such a good job that companies building coding agents opt to build on top instead of reinventing the undifferentiated parts
it's tedious - getting things working across hundreds of LLM providers, dealing with quirks, collaborating with all the companies involved to improve things
and then the platform work to get things working well on all environments
thankfully we have a solid business model and can devote significant resources to doing all of this - including funding things like OpenTUI
the forks that have popped up are confirmation we're starting to hit these goals so that's great - even when they're from competitors
what's less great is they seem bent on passing off all this work as their own, even claiming they "built it from scratch"
they've never communicated with us, never privately said "hey thanks for the work"
every company can choose how they want to operate but this isn't very aligned with the spirit of open source and these small gestures of good will go a long way
NeetCode retweeted
i've always been pro making money, being money motivated
and if i tried hard to do something great then money would be a side effect
it feels like silicon valley lately is infected by a different version of this where the money is the entire point
very uninspiring
NeetCode retweeted
Because if u build a system around say opencode or some other generic core, even if you're using Anthropic models today, you're a one line change away from moving to Gemini or OpenAI or whatever
What they're trying to do is force you to build yourself into a walled garden
NeetCode retweeted
I actually think more about the skill floor, which I think is incredibly high. These tools just are not that useful for building anything even remotely interesting unless you actually know how to code
As an example a few days ago I did a big refactor on btca (the actual pr, still finishing it up: github.com/davis7dotsh/bette…) that I pretty much entirely used opus for.
The thing is I spent about 45 minutes going back and forth writing a plan and md files with opus before it wrote a single line of code. And before I even started talk to it I had a plan of my own with scratch work functions and the full list of new primitives/how they work. It would have been impossible for me to design the new architecture or answer any of opus' questions if I didn't know how to code
The end result is ~4 days of work in ~2 hours, but that's only possible because I could have done the 4 days of work if I didn't have access to the AI.
As for the ceiling I think it's absurdly high and we haven't even scratched the surface, but that's just speculation. (and will be entirely based on already being a good programmer, just faster)
