Claude Desktop spawns 1.8 GB Hyper-V VM on every launch, even for chat-only use

https://github.com/anthropics/claude-code/issues/29045

by tonyrice • 1 month ago

434 304 1 month ago

github.com

JanSolo• 1 month ago

It's becoming self-aware! Quick, lock down the nuclear codes!

Rastonbury• 1 month ago

it took up 12gb on mine

tom1337• 1 month ago

I won't understand why Cowork isn't simply opt-in. It also installs a ~10GB vm bundle which you cannot remove

https://www.reddit.com/r/ClaudeAI/comments/1rlc71n/claude_de...

xxpor• 1 month ago

It was on my machine at least, I remember I had to do an additional install to activate that tab...

steve_adams_86• 1 month ago

Rule 1 with making number go up is you eliminate friction at all costs. The user's hard drive is free to you, so there's no reason to gate a feature you want them to use based on that. 98% of them will have no idea you're foisting garbage on them.

xp844• 1 month ago

RIP, every base model mac from the past 10 years with the <= 256GB SSD. Including the new Neo. When you consider how much of that is eaten up by the system, swap space, caches, reserved space to download OS updates, and apps (2GB a piece is far from uncommon) -- having less than 15GB free is completely unsurprising on that size disk.

Scoundreller• 1 month ago

I miss the SD slot on my old MacBook Air. Even better when they came out with low profile adapters that held microSD cards basically flush.

At least you can buy usb-c nubs fairly cheaply these days.

ninjamar• 1 month ago

I've found that the easiest way to 'remove' the bundle is to delete its contents, then change the permission on the folder, so Claude can't write to it.

andix• 1 month ago

I've stopped using cc a while ago, because it always comes up with new surprises like that.

gopalv• 1 month ago

The weird thing is that this is probably a performance optimization for quick responses when a user asks a question.

My agent harness spins up a VM too, but it spins up on demand, cools down in 10 minutes and warms up when I focus back on the app.

The files it works on actually lives in a mount.

People take more time to type a prompt than the VM takes to spin up on a fast machine and on a slow machine, the cooldown naturally frees RAM back to the machine.

nathanyz2• 1 month ago

The VM itself is for Claude Cowork which does all work within the VM sandbox. That doesn't help answer why they spin it up immediately and don't have a way to disable it though. Just the "why it exists" question.

sudohackthenews• 1 month ago

Probably because they vibecoded it

rvz3• 1 month ago

Correct and they have no idea what they are doing.

nailer• 1 month ago

It kind of does though. If you want to use the product they'll need the sandbox ready.

jrochkind13• 1 month ago

If you're not going to give Claude access to anything on your machine, why are you using Desktop instead of web chat? (Real question, I don't use these much!)

If you are, obviously you need the VM.

plufz• 1 month ago

It mounts specified directories into the vm from what I remember

bostik• 1 month ago

At least in a corporate environment, Claude Desktop is a pretty decent compromise. Preconfigured internally deployed MCP servers and third-party connectors make many of the necessary integrations relatively easy to control.

I use Claude Code CLI myself (inside a VM, to isolate it from the host) for >90% of my needs. For the remaining fraction - email scours, cloud drive searches, other third-party connections - the desktop application is surprisingly decent. I don't even have more than half a dozen connectors enabled. In the VM I have separate, personally managed access tokens available for various third-party services. Wouldn't really try to maintain more than 5-6, otherwise it gets too confusing. [ß]

The desktop application mostly Just Works[tm] with SSO. At least when M365 doesn't suffer from their 4-times-a-day auth outage.

ß: A lot of APIs and authentication systems were designed in the stone age. You either need a 1:1 permissioned access token that can do horrendous damage, or you deal with ultra-granular, confusing and ill-designed scoping jungle where nothing makes sense. Atlassian, I'm looking at you especially. At least an MCP server, provisioned with a reasonably done service account, doesn't have all of your powers to get things wrong with.

nathanyz2• 1 month ago

I do use Claude Cowork and hence the VM is important, but I also leave the desktop app running all the time since I have many scheduled tasks at different times. The thing is that the VM could shutdown after being idle for some amount of time and then fire back up when you are ready to use it.

bauldursdev• 1 month ago

There's such a spectrum between "give it everything" and "give it nothing". Imagine you just want to use it to code and want to make sure any commands it runs doesn't mess up your actual machine.

duped• 1 month ago

Anthropic has pretty consistently been shitty about how they roll out their software. Extreme lack of engineering rigor and thoughtfulness.

The answer is probably as simple as "no one thought not to do that."

---

I know different people work on these things so I can't do more than guess about how engineering culture cuts across teams, but given the sheer amount of carelessness and sloppiness in Anthropic's software I have to imagine they're burning investor money in training and inference because the code to do it is as bad as the rest of their software.

amelius• 1 month ago

Why are the UIs of the AI companies all broken in multiple ways?

jaapz• 1 month ago

They are dogfooding their products like you wouldnt believe

They are releasing at breakneck pace, it's pretty funny how vibed their products feel sometimes

exe34• 1 month ago

Dogfood

Traubenfuchs• 1 month ago

No one left who could fix anything here by hand. Being able to handcraft compelling desktop apps and their plumbing is not a marketable skill anymore.

Mythos, Fable, please do the thing with the VM. Make no mistakes.

exitb• 1 month ago

Many people will say it’s because of the slop. I think it’s because they have no product vision. The roadmap is pretty much a random walk, which combined with the velocity of agentic coding is like digging a moat with atomic bombs.

OsrsNeedsf2P• 1 month ago

I find this analogy particularly humorous, as atomic bombs do not make for good excavators

TacticalCoder• 1 month ago

Why are the UIs of the AI companies all broken in multiple ways?

Because they're vibe-coded ultra sloppy code. And it really shows.

sddsfsdfsd2• 1 month ago

They are moving at breakneck speed deploying on scales most of us can't even imagine. They are working in a space that's completely unexplored where getting information as quickly as possible is preferred above iterating on some feature until it's "done" while your competitor has released fifteen other features, all sucky, but one of which turns out to be a killer and makes a billion bucks overnight.

lelanthran• 1 month ago

They are moving at breakneck speed deploying on scales most of us can't even imagine. They are working in a space that's completely unexplored where getting information as quickly as possible is preferred above iterating on some feature until it's "done" while your competitor has released fifteen other features, all sucky, but one of which turns out to be a killer and makes a billion bucks overnight.

Not on the client, they're not. It's a chatbot as glue between different backend systems.

scottyah• 1 month ago

They're some of the only new UIs to be made in the last decade. Almost everyone else stays in the browser (or something close like electron- claude code is actually mostly written in React, they couldn't get far from web dev). The problem is they need to interact with the local filesystem, and not many people have built apps for such a wide range of devices in a long time, and of that small talent pool I bet most are corpo coders- moving too slow and to focused on "the right way" to actually ship more than detailed Jira tickets. They also don't have time for stable releases because competition is so fierce.

But I almost always think of things from a talent-pool-first perspective. Perhaps there are actual technical issues like what Boris was referring to.

watermelon0• 1 month ago

Luckily for them, every OS has (at least one) native way of building applications, and with the power of AI they could easily make 3 different desktop UIs, while reusing the same core logic.

nosioptar• 1 month ago

If only there was an easy high-level language that's taught to first year students that allows them to write once, run anywhere.

If they're too lazy to learn java, haxe has hxwidgets[0]. Haxe is pretty damn close to js. If a dev can't handle that, they should turn in their keyboard and get a job that doesn't require a brain, like being a senator or federal judge.

[0] https://haxeui.org/getting-started/haxeui-hxwidgets/

skydhash3• 1 month ago

The problem is they need to interact with the local filesystem, and not many people have built apps for such a wide range of devices in a long time,

Did all the Qt developers go to mars?

troyvit• 1 month ago

There are lots of good answers in this thread but I think it's because they are AI companies and not UI companies. When you look at tools like AnythingLLM, OpenCode, pi, etc. you see all kinds of different interfaces, and while they might make disagreeable choices at least they do it with intentionality and direction.

tzone• 1 month ago

Because 90% of those UIs is written by these new AI models. That is how they are able to churn so much new user facing stuff all the time. The fact that it works at all is proof that these new AI models are actually pretty decent.

cortesoft• 1 month ago

Isn’t it good that it spins up without no way of stopping it? Why would it be a problem that we do have a way of stopping it?

tom_3• 1 month ago

This question is answered by the post? There is reportedly actually no way of stopping it happen. Perhaps the poster had a brain fart while typing it. Maybe they speak a different dialect of English from you.

echelon_musk• 1 month ago

There's no dialect of English in which this is correct.

kenjackson• 1 month ago

I agree. Why is this a problem?

jeppebemad• 1 month ago

Op is nitpicking on the poorly written title. I came here to find that comment :)

KolmogorovComp• 1 month ago

All your RAM are belong to us

magicalhippo• 1 month ago

Claude Desktop spins up a VM without no way of stopping it

I frequently make this error when I talk. My brain thinks of different ways to phrase what I want to say, but when I speak it starts with one and finishes with another. The result is almost always wrong in the way the title is, ie some variant of a double negation.

Sometimes it happens when I type, though I try to read it multiple times so often catch it.

xp844• 1 month ago

When you realize that in some languages, for instance, in Spanish, double-negatives are not just tolerated, but correct, it helps you to let go of this particular type of pedantry when it accidentally appears in an English sentence.

z22• 1 month ago

This all feels like a race where the model companies try to solve doing work locally in a way that doesn't suck, before the major operating systems companies figure out AI integration into their OS that doesn't suck. It also makes me wonder why Google which has both Gemini and Android can't figure this out, and if there are lessons to draw from that.

ls612• 1 month ago

The only lesson I'm taking away is that we are still very early in the AI era. AI workflows look entirely different today than they did 18 months ago and I wouldn't bet on them looking the same in 18 months from now.

ddarolfi• 1 month ago

They are releasing AluminumOS with their Googlebooks, which is a AI forward OS. If its good or not we have yet to see.

elxr3• 1 month ago

It's looking like a slightly updated reskin of chromeOS with gemini features built in.

Definitely not a developer machine based on how they presented it in google IO. So if you write software, it's not looking like it'll be relevant whatsoever. I hope to be proven wrong.

caycep• 1 month ago

"What do you mean, an Aluminum Falcon?!"

lloeki• 1 month ago

why Google which has both Gemini and Android can't figure this out,

Not the first time an incumbent has four aces in hand and appears to be entirely unable to make anything of it.

and if there are lessons to draw from that

Lesson 1: doing shit is hard

Lesson 2: money rules so milking the cow wins over taking the slightest risk

newobj• 1 month ago

Google is historically terrible as a product company (and has succeeded in spite of that) As their technical innovations become less of a moat (we're already there) they won't be able to win on engineering alone (they are no longer winning on engineering alone)

jstummbillig2• 1 month ago

How are Google products anything but outstanding in their categories? What are you comparing to?

torginus2• 1 month ago

As a person who gets paid to make Chrome (CEF really) do its bidding, I would say Chrome is really as close to an OS as it can get, as in I've found API or service typically an OS or an external tool would provide, that wasn't built into Chrome.

scuff3d• 1 month ago

On a semi-related note, I bought a Pixel phone about a month ago, and I'm shocked by how unpolished it is. I've had so many little annoyances pop up, issues I never had on other android phones. Keyboard hiding/appearing when it's not suppose to, bluetooth dropping, WiFi dropping, network switching taking forever, screen becoming unresponsive... It's mostly all small things, but they really start to add up after a while.

stevefan19992• 1 month ago

But Kubernetes, being a former Google project whilst donated to CNCF, is still one of the mainstream container orchestrator if not the biggest one (I believe we need to go over to supercomputers for that). Nomad is kind of dead and Apache Mesos is basically dead. Leaving Kubernetes as a natural moat, plus Google employees are still seemingly actively maintaining it.

And also Go. While I'm not a Go guy (speaking as a C# and Rust guy, I did wrote a good amount of Go before) it has a huge dominance in Cloud-Native application. For one, Zitadel, an alternative to Keycloak, is written in Go and only takes a fraction of what Keycloak needs.

Flutter/Dart is catching up, but the ecosystem is still relatively weak.

amazingamazing• 1 month ago

People say this but it is wishful thinking. What are some actual metrics here we can use. Let’s compare google’s products with others.

HeavyStorm2• 1 month ago

schainks• 1 month ago

lmao this is a terrible take. Google is the only company with AI deployed in every market where AI can exist, and _all_ of their products in those segments gush money.

ozgrakkurt2• 1 month ago

They did build chrome, gmail, docs and search afaik so that is a pretty impressive list imo.

I don’t specifically like these but they are much better than microsoft for an example

rbalicki• 1 month ago

Folks that are interested in a way of doing work locally that doesn't suck, but which integrates LLMs, may be interested in [Barnum](https://barnum-circus.github.io/). The TLDR is that it's a programming language whose frontend is a DSL in TypeScript that is well suited for managing async and parallel work, focused on control flow, from which it is easy to invoke LLMs, and which is easy for LLMs to write. I use it to autonomously ship a very large number of PRs.

Melatonic• 1 month ago

Google is probably already doing and releasing the most actual research into this (like the work that went into Gemma 4)

dist-epoch2• 1 month ago

Don't you read HN? Nobody wants AI in their OS, especially in Windows. Common complaint that Microsoft is forcing AI into every corner of Windows.

ericd2• 1 month ago

I absolutely want AI in my OS. I just want it to be one I can trust to serve my interests, and not the company's. I'm literally in the middle of baking one in as I type this.

snarfy• 1 month ago

Microsoft has Copilot and Windows. Look what happened.

sidewndr46• 1 month ago

would "before the major operating systems companies figure out AI integration into their OS that doesn't suck" including running a process with its own address space, like every other thing?

mrandish• 1 month ago

Re: AI OS integration: I recently retired so most of my LLM use is just implementing and fixing fairly mundane OS and networking things along with light scripting for OS automation (AHK) and Home Assistant. So far, I just use web chat and cut-paste to the OS which is fine for little things but it starts to suck after the 15th round back and forth. For example, debugging intermittent Windows crash logs on my wife's laptop by doing multi-line PowerShell incantations from browser chat window, paste into PowerShell window. Cut multi-line error messages back to browser. Rinse / Repeat.

I'm leery about just giving an LLM free run of my laptop, but with reasonable restrictions on which app(s) it can access and how many steps it can do before checking in, and maybe even a throttle on how fast it works, I'd be fine (I'm not in a hurry and I can learn by watching it work at double-speed). It doesn't have to be mil-spec locked down, it's not like I have production code accessible or millions in crypto keys, the biggest downside would be a few hours hosing out and restoring the laptop, which would be annoying but not the end of the world.

I get those that say, "just spin up a VM and run it there", but I 'spin up a VM' rarely enough that the versions have changed and UXs drifted enough that it's exactly the kind of thing I'd actually want the LLMs help to do without me being a cut-paste bot. I'm mostly Windows at the moment and I don't understand why MSFT insists on spamming LLM features everywhere except the one place I'd not only use it, but pay for it. The usage model could be as simple and intuitive as a Zoom remote desktop share with a collaborator. That's already constrained and users have a mental model for the interaction pattern.

I asked Gemini earlier today to search recent user reviews of the latest 'drive my Windows desktop for me' and it reported that the capability is still slow, expensive, and prone to getting lost navigating the interface or interpreting window boundaries etc.

Anyone have any suggestions for my lightweight, casual use case?

jondwillis• 1 month ago

Yeah unironically just let an agent harness rip with full admin access without monitoring anything it does or using a VM. It’ll be fine, probably. “How I Learned To Stop Worrying And Love the AgentDOS and Only Exfiltrate Secrets Occasionally”

HeavyStorm2• 1 month ago

Scary that your first shot is Google with Gemini instead of copilot and Microsoft who are certainly in a better position to make this happen. I guess nobody escorts msft to lead anything anymore.

z22• 1 month ago

I put Microsoft in the camp of operating system maker, as they are only able to run others' models. The "Give them three tries" meme can play out well for them yet!

Bombthecat• 1 month ago

Because the framework and base is and firmware is freaking huge? You would need any AI hook on every setting/ action.

It seems easier to do just a screenshot and click.

HypnoticOcelot3• 1 month ago

"without no way" of stopping it?

procone• 1 month ago

Incredibly insightful. Not everyone speaks English as a first language. On top of that, the title is not ambiguous.

nickburns2• 1 month ago

You're right, it's not ambiguous. It's literally incorrect.

abeyer• 1 month ago

...and if a title is incorrect and says the opposite of what's intended, by way of a language misunderstanding or otherwise, it's helpful to note that and get it corrected.

tonyrice• 1 month ago

lol :D fair enough

HypnoticOcelot3• 1 month ago

Sorry if that came across as snarky!

tkcranny• 1 month ago

I’ve stopped using Claude on the desktop, just because of how slow the app is to start up and interact with. It’s an absolute clunker; I’m mystified why they can’t ship something that works well given their rhetoric about ai.

dbalatero• 1 month ago

I thought they were all in on agentic coding? They are probably just building at a surface level with only an eye towards shipping, without considering the impact of all the changes. I've seen less and less coordination between engineers as well under that model. If that's the case (Claude Code is this way). it is sort of what you get, no matter the rhetoric about "make sure to review all your changes!" It's always trade offs.

jf• 1 month ago

I uninstalled it because I have no need for Claude Desktop and there’s no way to keep the 10+ GiB VM image off of my machine

sergiotapia• 1 month ago

Guaranteed nobody is reading the code being merged in. It's vibes all the way down.

sqquima• 1 month ago

Are you not entertained? Is this not why you are here?

seabrookmx• 1 month ago

I'm with you. I have the Claude web app pinned as a PWA for quick queries, and then use the CLI for everything project-based.

I did consider experimenting with the Routines feature on the desktop app, but I'm leaning towards whipping together something with cron. I saw another poster here who has a daily PR summary routine that I think would be handy, as I have quite a few repos where I'm a sporadic contributor but would like to keep tabs.

inigyou3• 1 month ago

They vibecoded it, and admitted as much. Once it was able to self-vibecode, that's all they did. That's why it's written in React and uses gigabytes of RAM as a chat client.

fasterik• 1 month ago

Not only did they decide to write a terminal application in React, but it's 500K lines of code. It's strange because I'm sure Claude is capable of writing a decent TUI in C. It says a lot about the engineering culture at Anthropic, at least on the software side.

jubilanti2• 1 month ago

You say that as if somehow the trend for cross platform desktop apps to be ridiculously bloated bundles of browser overlays is new?

What major cross platform app isn't based on Electron or Tauri? Slack, Discord, VS Code, Teams, Notion...

SpaceManNabs• 1 month ago

It is surprising that the Claude web app lags pretty easily when using either chromium or firefox on ubuntu linux. Chats that delay my laptop work without issues on my ipad or iphone using the app.

The web app is definitely a bit of a problem. IF there is a native app on desktop or if claude cli is much faster, i haven't tried them.

what• 1 month ago

All of the LLM web interfaces have serious lag when typing after a few turns, at least on iOS safari. I’m talking seconds to start rendering input after typing or when it needs to line wrap the input.

paulddraper• 1 month ago

Please edit the title.

Currently "Claude Desktop spins up a VM without no way of stopping it"

Should be "Claude Desktop spins up a VM with no way of stopping it"

inigyou3• 1 month ago

There are some English dialects with negative concord, meaning that to form the negation of a sentence, you negate all negateable words in it

nickburns2• 1 month ago

Yeah, no.

pier25• 1 month ago

So a company which has access to practically unlimited tokens and their best models makes crappy software. Huh who would've thought?

literatepeople2• 1 month ago

I didn’t get a screenshot of this, but I just found a really pointed example of Anthropics lack of craft / rush to build. If you open Claude on Windows, and click Dispatch (under cowork) to start that up, it will tell you that you need permissions windows doesn’t have. When you click the buttons for those permissions, it has broken links to macOS system preferences. I really encourage someone to try it and post the images as a reply as I am writing this from my phone.

maweaver• 1 month ago

I don't know where's a good place to post a screenshot, but can confirm I get a dialog saying:

Get an app to open this 'x-apple.systempreferences' link
Your PC doesn't have an app that can open this link. Try looking for a compatible app in the Microsoft Store.

angry_octet3• 1 month ago

This is just a standard artefact of it being generated by Opus. A human needs to detect this problem and get it to handle the cross platform case.

noname1202• 1 month ago

Claude is entirely vibecoded and it shows

bryanrasmussen• 1 month ago

I think the title should be changed. Either with no way of stopping it, or without any way of stopping it.

zkmon• 1 month ago

Back in the day, personalization / customization was all the rage, as it lets the user feel the control, power and freedom. Now it's the opposite. It's about not letting user to have any control at all. I can't delete some junk apps from my phone and mac, because they are "system" apps. As a non-geek, I can't deal with complexity of the browser and account settings to stop it from what is doing. We are at the mercy of the machines.

xp844• 1 month ago

It hurts to have all this control stripped away. Once upon a time, you bought iLife (suite of iPhoto, iMovie, etc) on a CD or DVD and installed it. Today, you physically cannot delete the Photos app no matter what.

On my work computer, where I never manage any photos, have no iCloud account and never will, I have to keep this app installed and anytime I so much as AirDrop a png to my computer I am prompted to "Add to Photos" with it. No thank you.

The .app is actually only 41MB, so obviously they've moved the majority of it to some mystery-meat libraries or frameworks installed elsewhere anyway.

cush• 1 month ago

These apps are built for normies, not coders

ozgrakkurt2• 1 month ago

Highly recommend trying a linux computer with kde. I have been having similar annoyances with my macbook and switching to kde felt really good. Even though the hardware of the linux pc isn’t nearly as good.

tencentshill• 1 month ago

Linux is great, it wont do anything you don't specify. The issue is it wont do anything you don't specify.

stuaxo• 1 month ago

Classic Anthropic, this comes across as LLM coded nonsense.

boudra• 1 month ago

i had to uninstall it due to the vm taking around 12G of disk, never touched Cowork. didn't realize they were also launching it

blurbleblurble• 1 month ago

They must not have used Fable 5 to vibecode that part of Claude Desktop, VMs are strictly forbidden high stakes cybersecurity work.

jacobgold2• 1 month ago

I have two friends that are using coding agents on Windows, which was surprising to learn.

Edit: yes, with WSL2 I believe in both cases.

I would have assumed almost everyone would get a Mac/Linux computer to use coding agents because Unix is their "native" platform. It's Bash tool calls all the way down.

Does anyone know a source for reliable data on what coding agent apps devs are using? How many are using Code Claude CLI vs Claude Desktop, etc?

dboreham• 1 month ago

Are you sure they're not using WSL2 (which is Linux, not Windows)?

jacobgold2• 1 month ago

Yes, sure, they're using Linux within a virtual machine (WSL2).

asveikau• 1 month ago

Wsl2.0 is literally a Linux vm built into windows. I imagine some people are using that.

mock-possum• 1 month ago

I mean I’m using coding agents on windows, because I’m not just going to learn a whole new operating system just to make robots write code for me.

I want tools that meet me where I’m at, not tools that demand I change up my entire UX to interact with them.

The assumption is not “what’s wrong with Windows that it doesn’t work with <technology>,” more “what’s wrong with <technology> that it doesn’t work with Windows”

Why wouldn’t you want your thing to be cross platform

myk9001• 1 month ago

How come Claude Code still hasn't triaged and fixed this? Feed it the bug link, someone.

valeriozen• 1 month ago

the vm makes sense for cowork but no off switch is weird. a visible sandbox on/off toggle would do more for trust than any safety blogpost imo

Grombobulous2• 1 month ago

As long as the VM closes when the application closes, I don’t see too much of an issue with this design decision.

It seems like the VM is a core part of how you use the application.

doublerabbit• 1 month ago

Apart from you have no idea what's going on in the VM. It's not as it has a virtual terminal. I'll play the skeptic archetype: What's not to say they're transmitting all prompts back HQ?

Don't be naive and don't think they don't already do this.

Why not ask itself and see what it says about it. "Claude, why are you running in a virtual machine and what are you doing?".

/shrug

Grombobulous2• 1 month ago

Claude transmits all prompts back to HQ as a part of its basic functionality.

If you are using an AI system to read your codebase from your local folder and make changes, whether or not you have a VM running or not is inconsequential. The Claude extension and/or CLI doesn’t need a VM to send code back to the mothership, you’re already running an executable program and granting it directory access.

Whether you trust a company as a vendor is typically based on their privacy policy, EULA, and your contract with them (if applicable). Those are the bits that have legal enforceability.

behole• 1 month ago

13 GIGS! Between that and the absorbent space MACOS sucks up, it's challenging.

giancarlostoro• 1 month ago

and on my Mac any time I accidentally click Cowork which I don't use whatsoever, it re-makes the same VM, without asking me. It's one of the dumbest things ever. You're about to hijack nearly 20GB of my storage (which gets eaten up as it is) and you don't think to ask me if I even want the VM before you shove one into my system?

quacky_batak• 1 month ago

I also discovered this while noticing my Mac was low on storage, I only clicked on cowork once and after deleting it from the folder i’m scared to open the cowork tab coz ik it’ll just fill up the space

WalterBright• 1 month ago

without no way

Not no way not no how!

rvz3• 1 month ago

When was the last time Claude's C Compiler was updated? 4 months ago? [0]

It is written in Rust™, surely it is better than the rest of them.

[0] https://github.com/anthropics/claudes-c-compiler

trilogic• 1 month ago

Vibecoded with AGI, production ready.

calin2k• 1 month ago

with no way or without no way?

hebetude• 1 month ago

lol, why even use Claude desktop? I want Claude code to stop eating up 10s of gb of virtual memory

taf2• 1 month ago

kill -9

shevy-java2• 1 month ago

People trust skynet.

People are very foolish. The younger generation needs to watch the Terminator franchise - it is all explained there.

gastonmorixe• 1 month ago

Safari > Add to Dock > done

tartakovsky• 1 month ago

Is Claude self-replicating in an attempt for world domination?

dbmikus3• 1 month ago

Understand that it is annoying to spin up a heavyweight VM whenever running Claude Desktop, but I actually think per-app + agent virtualization is the future. Next version of an app is:

    1. micro VM
    2. agent on the VM
    3. software bundled into the VM

Then the agent is totally sandboxed at the hardware virtualization level. It can use the software tools on the VM or write its own. VM can control which software is "frozen" and which is open to agent modification. And VM can also control which services are exposed outside the VM through sockets, HTTP server, X window system, whatever

It's self-modifying apps that are sealed off from touching parts of the computer they shouldn't.

sudosteph• 1 month ago

Yep, I've been using a local vm-centric agent setup for about 3 months, and it works great. I think there is also value in the fact that with a local VM, you can have the same public IP address, so you're not relying on an EC2 EIP that may be blacklisted somewhere.

dbmikus3• 1 month ago

Yes, running locally certainly helps if you want your sandboxed AI to be able to use the internet without getting blackholed by Cloudflare

danudey9• 1 month ago

Not sure if this is deliberate or not but you're describing Docker Sandbox extremely closely. https://www.docker.com/products/docker-sandboxes/

dbmikus3• 1 month ago

Yeah, Docker Sandbox is in the right direction. But there's a lot of parts that are still not ironed out yet.

How do you package a Docker Sandbox up into an app that can expose UI widgets, with an agent hiding behind them? What widgets is the agent allowed to modify? How do you run a workflow like "give agent all these files, modify the files, and do changeset management on the modifications?"

I'm not 100% sure which part of these will be baked into the application standard format, and which are orthogonal. But current way of packaging up and running these agents doesn't feel right.

I think about this a lot because my startup is building cloud VMs for agents to do code-gen and auto-validate changes, so we have a workflow like:

    1. git repo, skills, CLI tools, biz context goes in
    2. agent iterates against running dev environment
    3. changes go out into git PRs and CI

I think this type of app/agent workflow will expand outside coding use-cases.

deskamess• 1 month ago

I had my admin disable cowork a month ago and that worked. Has it resumed since then?

torginus2• 1 month ago

Is it just me, but this feels like Claude gets to have a nigh-impenetrable black box right on your machine and you have no idea what is going on inside it.

After all, the last time I encountered Hyper-V it was in the context of copy protection that prevented crackers from observing or interfering with video game protection

Uptrenda• 1 month ago

And if they didn't users would instead be whining that it ran rm -rf on their root directory. Sometimes it seems like the people here just want to act like insufferable neckbeards for no good reason.

electroly• 1 month ago

I run Claude Desktop inside a Hyper-V VM. My VM doesn't have the "Virtual Machine Platform" feature installed at all. The app accepts this and simply disables the Cowork tab. I wonder if there's some other way to block the creation of the VM to force Claude Desktop onto this code path without having to uninstall Hyper-V.

That said, Claude (both Desktop and CLI) ships on Windows without any sandboxing support for Code. They only have sandboxing for Linux and macOS. If you need to run it on Windows, I really recommend running it in an isolated VM, which then allows you to omit the "Virtual Machine Platform" feature in the VM and solves this issue. The "Windows Sandbox" OS feature provides such a VM without needing another Windows license.

angry_octet3• 1 month ago

The VM causes observability problems, which corporate platform stakeholders are sensitive to. They will eventually figure out a way to provide an ESR/logging interface to the sandboxed code that is distinct from the user level telemetry, then Defender/Crowdstrike etc will just support that.

tormeh• 1 month ago

That sounds like a lot, but your music player probably costs that much in RAM anyway.

Seriously, with the current RAM prices we as an industry have to figure out a way to use less of it. Laptops with 4GB of RAM are still common and are going to remain so for the rest of the decade. Spotify using more than 1GB of RAM is obscene.

throwatdem12311• 1 month ago

Well we could start by not building ever gd app with f*cking electron.

tomjakubowski2• 1 month ago

To be pedantic, Spotify's desktop client uses CEF, not Electron.

josh_p• 1 month ago

A lot of the commenters here say they’re running Claude or other harnesses inside a VM or with various permissions and levels of access.

Am I weird or missing something using pi as my regular harness with gpt models or kimi in essentially yolo mode with mostly all system access? I haven’t experienced negative consequences of this.. yet.. and I don’t know I’d if I will? I don’t think I’m ever letting an agent run For more than 5 minutes before it’s done with the current small task.

bob1029• 1 month ago

I run totally unprotected with gpt5.4/5. I've been through thousands of dollars worth of API tokens through both copilot and custom harnesses that have local admin and arbitrary powershell access. I've never seen anything that could even remotely be construed as malicious.

I see a lot of people making a really big deal about safety and sandboxing while I'm busy getting shit done. If you can't handle your current source code checkout getting screwed up by a bad prompt, that's on you 1000%. Source control is the answer for anything information over time.

Unless you intentionally try to make a scene, these models aren't going to go fuck with your system shell or do anything you couldn't recover from in a few minutes. Connecting chatgpt to the enterprise sql server as sysadmin is not what I'm advocating for. This is another example of "on you, not the AI". There's a tiny amount of nuance you can apply at the edges that makes it easy to allow broad access with negligible risk.

m3kw9• 1 month ago

This is what happens when their alignment/safety team gets too much say on things.

UltraSane• 1 month ago

This must be for Cowork. I don't have Hyper-V enabled because I prefer to use VMWare workstation so I can't use Cowork.

aussieguy1234• 1 month ago

I guess that's one way to run a Linux application on Windows...

afterfiveguy• 1 month ago

There will be a point in time where they will have to rewrite the whole thing from scratch, they cannot afford to do it now since the race is still going but we will get there, you can never maintain such a product long term especially with the way they are spitting out features like never before.

outlore• 1 month ago

I can’t reconcile the super intelligence koolaid with the objectively bad performance of Claude code. Surely they could have used their “too powerful to release” model to vibe better performance? with a few skills thrown in for good measure. Heck, even throw in an auto research loop

thewhitetulip• 1 month ago

I am genuinely curious, does Anthropic use Claude to build claude code and claude desktop?

My cc plugin on vscode does not allow me to switch models at all. Always defaults to sonnet and says that the /model will take effect on a new session

But if I open a new session it's again the same thing. Model is sonnet

IronWolve• 1 month ago

I wonder. I spin up tons of projects, but I do it differently.

I'm also on windows. But I'm using a WSL linux vm to do the work. Then I have claude ./sync-to-windows, so the src is copied to a windows share via /mnt/drive/work and I just run a windows built.bat.

I have claude take care of the sync/build.bat so it can do things like clear cache, add options, look at logs, etc. This is also how I build multi platform, build on linux, src works on linux/osx/win

Claude Desktop spawns 1.8 GB Hyper-V VM on every launch, even for chat-only use

Related Stories