Python and PyTorch all call out to C libraries… I don’t get what he means by “proving LLMs can run without Python and PyTorch” at all. Seems like they don’t understand basic fundamentals about things here…

All because this dude is the ultimate judge for all that is good and worth doing somehow..

Oh.. So you start doing something new and you're top 10% without practicing or being bad at it first? I'd love to test that to see if it's the case... Your logic is "You're not the best ever to do something so you are not doing it" means you have probably never done a single thing your entire life. Maybe you should just stop.

Obviously familiar with Fabrice Bellard and his technical contributions but it seems like he is a pretty private person and he keeps to himself. I don't really know much about him as a person.

As someone who does it for a living the challenge can be in both. However this article is asking its agents to do CTF like challenges which I am sure the respective LLMs have seen millions of so it can essentially regurgitate a large part of the exploit code. This is especially true for the OOB/RW primitive API.

Yea, this entire repo/article seems super misleading to me. Not to mention asking it to generate API for OOB R/W primitives is essentially asking it to regurgitate what exists on thousands of github repos and CTF toolkits.

Most modern kill chains involve chaining together that many bugs... I know because it's my job and its become demoralizing.

Vulnerability Researcher/Reverse Eng here... Aspects about it generating an API for read/write primitives are simply it regurgitating tons of APIs that exist already. Its still cool, but its not like it invented the primitives or any novel technique. Also, this toy JS is similar to binaries you'd find in a CTF. Of course it will be able to solve majority of those. I am curious though.. Latest OpenAI models don't seem to want to generate any real exploit code. Is there a prompt jail break or something being used here?

I had similar questions when reading the original article. I’m also interested in how the agent is constructed. From my experience, it can be very difficult to implement exploits without access to debugging tools, so I’m curious whether pwndbg or similar tools are included in the agent’s toolset and, if so, how they are integrated. Existing open-source GDB MCPs don’t work very well unless further optimized, at least the last time I checked.

It is pretty and it can do pretty much exactly what Rust enums do if they learned basic idiomatic Go.. Rust is a cult at this point honestly.

No its because 99% of the time people use enums to give names to magic constants... That is it. Go went for simplicity and const+iota achieves it just fine. People act like enums make or break software itself or something.

That seems unlikely to me to be the actual explanation. It could very well be what you prefer or how you would do it, but I can definitely assure you that the Go/other infrastructure teams think about these problems and hear plenty of complaints about lack of union type support.

Yea but with a tiny bit more effort they could have ensured that an invalid value is never assigned to an enum, iterate over the values, ensure switch statements handle every case, etc.