Thanks for posting! ~15k lines of code, a lot to poke around in. I was working on a Siemens S7 PLC project with a WINCC HMI for a power plant (the same target of the cyber-weapon) as I listened to the audio book[2] based on this ~12 years ago, entirely changed how I viewed critical industrial infrastructure. One quote from the book that stuck with me was how you can only use a cyber weapon once at full potential, as it’ll either get patched and/or everyone can reverse engineer it to use.
For those not familiar with Stuxnet, it’s a discovered cyber-weapon from 2010 which “reportedly destroyed almost one-fifth of Iran's nuclear centrifuges. ” and “ neither the United States nor Israel has openly admitted responsibility” but likely were the developers [1]
I always enjoyed Kim Zetter's work. Well written and researched, and she doesn't put herself in the story, unlike some other authors. She highlights the practitioners who are actually the ones doing the work.
I always wondered about how feasible the usb drive propagation bit always noted was. Was there ever any evidence that the hardware was already infected at a less scrupulous reseller? I’ve heard of another site in EU that had misbehaving s7. It could have been a reseller that played loose with licensing.
This looks like slop, it's all concatenated into a single file and most probably not based on the actual malware. I'm fairly sure that for example the real one does not include the literal string "Stuxnet" anywhere, like it does here:
The original name given by VirusBlokAda was "Rootkit.Tmphider;"[41]
Symantec, however, called it "W32.Temphid", later changing it to "W32.Stuxnet".[42]
Its current name is derived from a combination of keywords found in the software (".stub" and "mrxnet.sys").[43][44]
Directory filtering needs to be fixed, one weird filename or symlink will make it BSOD.
SSDT should probably have a lock. The chance of a race is ~low (higher under heavy sustained workloads) but it's too important to leave to chance.
I'd probably do a rebuild of the directory lists in a separate buffer instead of working in place to avoid alignment fuckups.
Yes I used LLMs, just like I did for all of the other vulns I've found or refined. As you can see from the source, this shit is tedious as hell. Doesn't change the value of knowing what to look/ask for.
Give one of those open models a fresh windows box (not a VM) and tell it to fuck something up, it's fun.
If you're using coding agents for this, it may be worth splitting this up into multiple well arranged modules that tell a coherent story and make it easy to browse, and add explanatory docs based on the various things the LLM has found about each function / type.
Your decompilation threads have all the necessary info in them for this, anyone coming after lacks that foundation and effectively is doing a second inference over the hidden state, assumptions, etc. that your sessions have in them. A simulacrum of a simulacrum in essence is likely to be not particularly good.
Thanks for posting! ~15k lines of code, a lot to poke around in. I was working on a Siemens S7 PLC project with a WINCC HMI for a power plant (the same target of the cyber-weapon) as I listened to the audio book[2] based on this ~12 years ago, entirely changed how I viewed critical industrial infrastructure. One quote from the book that stuck with me was how you can only use a cyber weapon once at full potential, as it’ll either get patched and/or everyone can reverse engineer it to use.
For those not familiar with Stuxnet, it’s a discovered cyber-weapon from 2010 which “reportedly destroyed almost one-fifth of Iran's nuclear centrifuges. ” and “ neither the United States nor Israel has openly admitted responsibility” but likely were the developers [1]
[1-Wikipedia Entry](https://en.wikipedia.org/wiki/Stuxnet)
[2-“Countdown To Zero Day” book if you liked the Wikipedia entry](https://www.audible.com/pd/Countdown-to-Zero-Day-Audiobook/B...)
[3-“Zero Days” movie](https://www.imdb.com/title/tt5446858/)
This is amazingly awesome. Very intriguing and what a great idea in the first place. One of the most important pieces of modern software IMO.
If you're going to spend tokens on the RE for this, it would've been nice to spend some tokens documenting it or making it easier to navigate
I very much recommend reading "Countdown to Zero Day: Stuxnet and the Launch of the World's First Digital Weapon". Fascinating stuff.
I always enjoyed Kim Zetter's work. Well written and researched, and she doesn't put herself in the story, unlike some other authors. She highlights the practitioners who are actually the ones doing the work.
I always wondered about how feasible the usb drive propagation bit always noted was. Was there ever any evidence that the hardware was already infected at a less scrupulous reseller? I’ve heard of another site in EU that had misbehaving s7. It could have been a reseller that played loose with licensing.
This looks like slop, it's all concatenated into a single file and most probably not based on the actual malware. I'm fairly sure that for example the real one does not include the literal string "Stuxnet" anywhere, like it does here:
Wikipedia about the origin of the name:Yeah... it's reverse engineered. They explained that.
Astra can reverse engineer binaries now. That's likely where it came from.
But then did it hallucinate that registry key? Or this?
/seta centrifuge_speed 1000000
Somehow, I don't think the original authors of this proprietary code are going to be filing a copyright claim...
A NSL seems more likely.
It would be really amusing to see, though.
Secrecy aside, there may be some other issues with copywriting a product (allegedly) made by a government.
Directory filtering needs to be fixed, one weird filename or symlink will make it BSOD.
SSDT should probably have a lock. The chance of a race is ~low (higher under heavy sustained workloads) but it's too important to leave to chance.
I'd probably do a rebuild of the directory lists in a separate buffer instead of working in place to avoid alignment fuckups.
Yes I used LLMs, just like I did for all of the other vulns I've found or refined. As you can see from the source, this shit is tedious as hell. Doesn't change the value of knowing what to look/ask for.
Give one of those open models a fresh windows box (not a VM) and tell it to fuck something up, it's fun.
If you're using coding agents for this, it may be worth splitting this up into multiple well arranged modules that tell a coherent story and make it easy to browse, and add explanatory docs based on the various things the LLM has found about each function / type.
Waste of tokens IME, better to do it like this and leave that kind of post processing to signature obfuscation.
You don't need exploratory docs, try asking one for complete function signatures of all the Windows APIs used, it's going to spit the docs out.
Your decompilation threads have all the necessary info in them for this, anyone coming after lacks that foundation and effectively is doing a second inference over the hidden state, assumptions, etc. that your sessions have in them. A simulacrum of a simulacrum in essence is likely to be not particularly good.
waiting for Pegasus source code