I rushed out nono.sh (the opposite of yolo!) in response to this and its already negated a few gateway attacks.
It uses kernel-level security primitives (Landlock on Linux, Seatbelt on macOS) to create sandboxes where unauthorized operations are structurally impossible. API keys are also stored in apples secure enclave (or the kernel keyring in linux) , and injected at run time and zeroized from memory after use. There is also some blocking of destructive actions (rm -rf ~/)
its as simple to run as: nono run --profile openclaw -- openclaw gateway
You can also use it to sandbox things like npm install:
nono run --allow node_modules
--allow-file package.json package.lock npm install pkg
Its early in, there will be bugs! PR's welcome and all that!
lol thanks! seriously, I have been running the tool over and over while testing and I kept typing 'nano' and opening binaries in the text editor. Next minute I swearing my head off trying to close nano (and not vim!)
hedgehog 19 minutes ago [-]
Obviously I'm biased but this looks really useful.
krackers 58 minutes ago [-]
Is this better than using sandbox-exec (on mac) directly?
decodebytes 46 minutes ago [-]
Hmm, I don't know about better, more convenient I guess. But if it floats your boat you could write out everything in the sb format and call sandbox_exec()!
overgard 1 hours ago [-]
I'm curious, outside of AI enthusiasts have people found value with using Clawdbot, and if so, what are they doing with it? From my perspective it seems like the people legitimately busy enough that they actually need an AI assistant are also people with enough responsibilities that they have to be very careful about letting something act on their behalf with minimal supervision. It seems like that sort of person could probably afford to hire an administrative assistant anyway (a trustworthy one), or if it's for work they probably already have one.
On the other hand, the people most inclined to hand over access to everything to this bot also strike me as people without a lot to lose? I don't want to make an unfair characterization or anything, it just strikes me that handing over the keys to your entire life/identity is a lot more palatable if you don't have much to lose anyway?
Does it matter? Let them cook and get burned if they want to.
hahuhs 56 minutes ago [-]
[dead]
mh2266 54 minutes ago [-]
The whole premise of this thing seems to be that it has access to your email, web browser, messaging, and so on. That's what makes it, in theory, useful.
The prompt injection possibilities are incredibly obvious... the entire world has write access to your agent.
???????
Trufa 39 minutes ago [-]
It is very much fun! Chaotic and definitely dangerous but a fun little experiment of the boundaries.
It’s definitely not it it’s final form but it’s showing potential.
voxgen 14 minutes ago [-]
I'm working in AI, but I'd have made this anyway: Molty is my language learning accountability buddy. It crawls the web with a sandboxed subagent to find me interesting stuff to read in French and Japanese. It makes Anki flashcards for me. And it wraps it up by quizzing me on the day's reading in the evening.
All this is running on a cheap VPS, where the worst it has access to is the LLM and Discord API keys and AnkiWeb login.
mentalgear 1 hours ago [-]
Moltbot is a security nightmare, especially it's premise (tap into all your data sources) and the rapid uptake by inexperienced users makes it especially attractive for criminal networks.
g947o 34 minutes ago [-]
We'll all have a good laugh when looking back at this in a few years.
catlifeonmars 15 minutes ago [-]
Any customers of products built on this stuff, who have their SSNs, numbers, and other PII leaked will not be laughing. But hey, who cares about them?
avaer 47 minutes ago [-]
Yes, there are already several criminal networks operating on it (transparently). I guess some consider this a feature.
cal85 44 minutes ago [-]
How do you know this? Not disagreeing, just curious.
avaer 40 minutes ago [-]
The links have been posted to HN if you search.
https://moltroad.com/ comes to mind. The "top rated" on there describes itself as "trading in neural contraband".
That's in addition to all of the actual hijacking hacks that have been going on.
I'm not saying any of this is successful, but people are certainly trying.
chrisjj 49 minutes ago [-]
It's like a bank decided to open its systems to a bunch of students it hired off Fiverr.
jungfty 48 minutes ago [-]
[dead]
ethin 59 minutes ago [-]
Things like this are why I don't use AI agents like moltbot/openclaw. Security is just out the window with these things. It's like the last 50 years never happened.
voxgen 3 minutes ago [-]
It's not perfect but it does have a few opt-in security features: running all tools in a docker container with minimal mounts, requiring approvals for exec commands, specifying tools on an agent by agent basis so that the web agent can't see files and the files agent can't see the web, etc.
That said, I still don't trust it and have it quarantined in a VPS. It's still surprisingly useful even though it doesn't have access to anything that I value. Tell it to do something and it'll find a way!
avaer 43 minutes ago [-]
No need to look back 50 years, people already forgot 2021 crypto security lapses that collectively cost billions. Or maybe the target audience here just doesn't care.
dotancohen 1 hours ago [-]
The real problem is that there is nothing novel here. Variants of this type of attack were clear from the beginning.
lxgr 1 hours ago [-]
What I would have expected is prompt injection or other methods to get the agent to do something its user doesn't want it to, not regular "classical" attacks.
At least currently, I don't think we have good ways of preventing the former, but the latter should be possible to avoid.
ethin 54 minutes ago [-]
They are easy to avoid if you actually give a damn. Unfortunately, people who create these things don't, assuming they even know what even half of these attacks are in the first place. They just want to pump out something now now now and the mindset is "we'll figure out all the problems later, I want my cake now now now now!" Maximum velocity! Full throttle!
It's just as bad as a lot of the vibe-coders I've seen. I literally saw this vibe-coder who created an app without even knowing what they wanted to create (as in, what it would do), and the AI they were using to vibe-code literally handwrote a PE parser to load DLLs instead of using LoadLibrary or delay loading. Which, really, is the natural consequence of giving someone access to software engineering tools when they don't know the first thing about it. Is that gatekeeping of a sort? Maybe, but I'd rather have that then "anyone can write software, and oh by the way this app reimplements wcslen in Rust because the vibe-coder had no idea what they were even doing".
lxgr 45 minutes ago [-]
> "we'll figure out all the problems later, I want my cake now now now now!" Maximum velocity! Full throttle!
That is indeed the point. Moltbot reminds me a lot of the demon core experiment(s): Laughably reckless in hindsight, but ultimately also an artifact of a time of massive scientific progress.
> Is that gatekeeping of a sort? Maybe, but I'd rather have that
Serious question: What do you gain from people not being able to vibe code?
hugey010 39 minutes ago [-]
Not who you're responding to, but I'm not a huge fan of vibe coding for 2 reasons: I don't want to use crappy software, and I don't want to inherit crappy software.
lxgr 19 minutes ago [-]
Same, but I've both used and inherited crappy software long before LLMs and agents were a thing.
I suppose it's going to be harder to identify obvious slop at a first glance, but fundamentally, what changes?
chrisjj 45 minutes ago [-]
> They just want to pump out something now now now
Some people actually fell for "move fast and break things".
ejcho 42 minutes ago [-]
I think with the advent of the AI gold rush, this is exactly the mentality that has proliferated throughout new AI startups.
Just ship anything and everything as fast as possible because all that matters is growth at all costs. Security is hard and it takes time, diligence, and effort and investors aren't going to be looking at the metric of "days without security incident" when flinging cash into your dumpster fire.
chrisjj 47 minutes ago [-]
> At least currently, I don't think we have good ways of preventing the former, but the latter should be possible to avoid.
Here's the thing. People who don't see a problem with the former obviously have no interest in addressing the latter.
bmit 1 hours ago [-]
So many people are giving keys to the kingdom to this thing. What is happening with humanity?
lxgr 57 minutes ago [-]
Humanity is the same it's always been. Some people are just inherently curious despite the obvious dangers.
Also, if you think about it, billions of people aren't running Moltbot at all.
hahuhs 54 minutes ago [-]
[dead]
ejcho 48 minutes ago [-]
do people even care about security anymore? I'll bet many consumers wouldn't even think twice about just giving full access to this thing (or any other flavor of the month AI agent product)
vulnwrecker5000 54 minutes ago [-]
what worries me here is that the entire personal AI agent product category is built on the premise of “connect me to all your data + give me execution.” At that point, the question isn’t “did they patch this RCE,” it’s more about what does a secure autonomous agent deployment even look like when its main feature is broad authority over all of someone's connected data?
Is the only real answer sandboxing + zero trust + treating agents as hostile by default? Or is this category fundamentally incompatible with least privilege?
So that's shifting the responsibility to users. And likely many users tools don't understand what those words mean.
All these companies/projects break decades of our security practice and sell you AI browser, AI agent for... I don't know what?
vulnwrecker6000 3 minutes ago [-]
Agents need fewer privileges!
chrisjj 51 minutes ago [-]
We need more Windows' "Are you sure you want XXX to make changes to your computer? (no I can't tell you what changes, but trust me.)"
/i
clawsyndicate 1 hours ago [-]
legit issue for local installs but this is why we run the hosted platform in gVisor. even with the exploit you're trapped in a sandbox with no access to the host node. we treat every container as hostile by default.
hughw 28 minutes ago [-]
You sound like the confident techie character in a Michael Crichton novel pronouncing "We've thought of everything there's no way for the demon to escape" shortly before the demon escapes.
chrisjj 41 minutes ago [-]
So... what use is an agent that cannot reach out of its trap?
electroglyph 1 hours ago [-]
that response is not comforting
nsm100 1 hours ago [-]
Thank you for doing this. I'm shocked that more people aren't thinking about security with respect to AI.
avaer 46 minutes ago [-]
People are thinking about it. I'm just not sure if the intersect between people who use OpenClaw/Moltbook is very high.
lxgr 56 minutes ago [-]
This isn't even AI security, as far as I can tell: It looks like regular old computer security to me.
g947o 12 minutes ago [-]
In the old days we just call that arbitrary code execution.
And these AI people just act as if that's never a problem.
Rendered at 22:27:53 GMT+0000 (Coordinated Universal Time) with Vercel.
It uses kernel-level security primitives (Landlock on Linux, Seatbelt on macOS) to create sandboxes where unauthorized operations are structurally impossible. API keys are also stored in apples secure enclave (or the kernel keyring in linux) , and injected at run time and zeroized from memory after use. There is also some blocking of destructive actions (rm -rf ~/)
its as simple to run as: nono run --profile openclaw -- openclaw gateway
You can also use it to sandbox things like npm install:
nono run --allow node_modules --allow-file package.json package.lock npm install pkg
Its early in, there will be bugs! PR's welcome and all that!
https://nono.sh
On the other hand, the people most inclined to hand over access to everything to this bot also strike me as people without a lot to lose? I don't want to make an unfair characterization or anything, it just strikes me that handing over the keys to your entire life/identity is a lot more palatable if you don't have much to lose anyway?
Am I missing something?
The prompt injection possibilities are incredibly obvious... the entire world has write access to your agent.
???????
It’s definitely not it it’s final form but it’s showing potential.
All this is running on a cheap VPS, where the worst it has access to is the LLM and Discord API keys and AnkiWeb login.
https://moltroad.com/ comes to mind. The "top rated" on there describes itself as "trading in neural contraband".
That's in addition to all of the actual hijacking hacks that have been going on.
I'm not saying any of this is successful, but people are certainly trying.
That said, I still don't trust it and have it quarantined in a VPS. It's still surprisingly useful even though it doesn't have access to anything that I value. Tell it to do something and it'll find a way!
At least currently, I don't think we have good ways of preventing the former, but the latter should be possible to avoid.
It's just as bad as a lot of the vibe-coders I've seen. I literally saw this vibe-coder who created an app without even knowing what they wanted to create (as in, what it would do), and the AI they were using to vibe-code literally handwrote a PE parser to load DLLs instead of using LoadLibrary or delay loading. Which, really, is the natural consequence of giving someone access to software engineering tools when they don't know the first thing about it. Is that gatekeeping of a sort? Maybe, but I'd rather have that then "anyone can write software, and oh by the way this app reimplements wcslen in Rust because the vibe-coder had no idea what they were even doing".
That is indeed the point. Moltbot reminds me a lot of the demon core experiment(s): Laughably reckless in hindsight, but ultimately also an artifact of a time of massive scientific progress.
> Is that gatekeeping of a sort? Maybe, but I'd rather have that
Serious question: What do you gain from people not being able to vibe code?
I suppose it's going to be harder to identify obvious slop at a first glance, but fundamentally, what changes?
Some people actually fell for "move fast and break things".
Just ship anything and everything as fast as possible because all that matters is growth at all costs. Security is hard and it takes time, diligence, and effort and investors aren't going to be looking at the metric of "days without security incident" when flinging cash into your dumpster fire.
Here's the thing. People who don't see a problem with the former obviously have no interest in addressing the latter.
Also, if you think about it, billions of people aren't running Moltbot at all.
Is the only real answer sandboxing + zero trust + treating agents as hostile by default? Or is this category fundamentally incompatible with least privilege?
yikes
no, they documented it
https://docs.openclaw.ai/gateway/security#node-execution-sys...
All these companies/projects break decades of our security practice and sell you AI browser, AI agent for... I don't know what?
/i
And these AI people just act as if that's never a problem.