binyu

7 ms·

181.
by binyu 4mo ago
Very cool work. What approach are you using? Been working on a similar in-browser node runtime based on Rust/WASM kernel + Service-Worker HTTP intercept + CJS→ESM transform. Feature wise, does this compare to StackBlitz webcontainers?
182.
by binyu 4mo ago
Been working on something similar based on Webcontainers. How does your Node.js support compare with StackBlitz technology? Are you running the version of Claude code that Anthropic distributes in the browser or did you have to adapt it to
183.
by binyu 4mo ago
Claude workflows in ultra code mode works in a very similar fashion and it consumes a moderate amount of the session usage limit, depending on the complexity of the task. With the API it would probably get expensive quickly though
184.
Anthropic's open-source framework for AI-powered vulnerability discovery(github.com)
539 points by binyu 4mo ago|142 comments
185.
by binyu 4mo ago
The Internet will be the Internet. Expect it to get worse if anything.
186.
by binyu 5mo ago
Yeah, you'd need to use asymmetric quantization and other software techniques.
187.
by binyu 5mo ago
> "exploit" More like social engineering meets AI and stupidity
188.
by binyu 5mo ago
Gotcha, I am not saying your setup is inherently wrong or useless. I am glad it works for your use cases. Godspeed
189.
by binyu 5mo ago
It depends but you cannot directly mix for example Ampere with Ada coz the lack of support for native FP8 in Ampere.
190.
by binyu 5mo ago
Yes but it creates a bottleneck that negates the benefit of using multiple cards that way. Look into it. Cheers
191.
by binyu 5mo ago
> Wait, wait, wait: browsers allow websites to store junk on my drive? Technically even a cookie is junk on your drive > Without even asking whether the site can use local storage? Would it be practical to ask permission for every sit
192.
by binyu 5mo ago
The V100 and the 4090 are based on vastly different architectures, the former uses the older Volta while the latter uses Ada. Last I checked you cannot meaningfully combine them. The 3090 is better than the V100, just get two 3090 and a NVL
193.
A Weekend in Claude Design Saves 3 Weeks of Claude Code(cashandcache.substack.com)
5 points by binyu 5mo ago|0 comments
194.
by binyu 5mo ago
MCP is what XML dreamed of becoming.
195.
by binyu 5mo ago
Truly makes me positive about the future. Thanks Andrej
196.
by binyu 5mo ago
> It makes individual agents untestable because their inputs and outputs are strings Strings can't be valid test vectors? Large language model are highly non-deterministic by design, no matter what.
197.
by binyu 5mo ago
It is much more of a context window size and model capabilities problem. Local models are not even remotely close in solving complex problems, even when used with the same harness.
198.
by binyu 5mo ago
That is true, it is a 1.6T parameters model so it requires a great deal of memory. I also heard there's a 2bit quantization that works well on Apple metal.
199.
by binyu 5mo ago
Right, thank you. Personally I think reading all the code that the AI produces is impossible and kind of defeats the purpose of using it. The key is to devise a structured way to interact with it (skills and similar) and use extensive testi
200.
by binyu 5mo ago
They all still fall short of Opus 4.6, definitely though. They are good but fail on extremely complex tasks, in contrast with a frontier model that will keep on trying until it succeeds or exhausts the solutions space.
201.
by binyu 5mo ago
DeepSeek V4 with 1 million token context window is pretty powerful, although still not there. There's hope that Opus 4.5 level performance locally is not that far away.
202.
by binyu 5mo ago
Gotcha, that implies you are going to read the code that the AI produces anyways. > Go reads fine whether the architecture is good or bad Were you reading the Golang code all along and got fooled or did you review it after it failed? Sor
203.
by binyu 5mo ago
> I'm rewriting k10s in Rust. Not because Rust is better but, because it's the language I can steer. I've written enough of it to feel when something's wrong before I can articulate why. That instinct is the one thing
204.
by binyu 5mo ago
Yeah, that part is probably not done by Claude.
205.
by binyu 5mo ago
I don't think this protects from sidechannel/timing attacks applied to images load time completely. Edit: Reading more thoroughly, probably it does to a great extent after all.
206.
by binyu 5mo ago
They forgot to add timing attack on images load time which can be used to tell if you visited X website. https://www.ieee-security.org/TC/SP2011/PAPERS/2011/paper010...
207.
by binyu 5mo ago
Hey, this is cool work. By any chance, did you see Cloudflare Artifacts? https://blog.cloudflare.com/artifacts-git-for-agents-beta/
208.
by binyu 5mo ago
Pinocchio had definitely some sociopathic traits as a character. The Pinocchio movie by Walt Disney from 1940 is a masterpiece. Highly recommended.
209.
by binyu 5mo ago
Very cool. Was this inspired by Cloudflare Artifacts? I'm working on a similar system as the backend for an upcoming agentic coding platform.
210.
by binyu 6mo ago
> which is easy, and more like changing the colour of a carbon-fibre Formula 1 part, which requires re-calculating the weight, strength and aerodynamics. Seems like a bit overstated for a CPU fan but I might be wrong

More ›