Cursor just got a major upgrade! Agents can onboard to your codebase, use a cloud computer to make changes, and send you a video demo of their finished work. The latency of using the remote desktop is smooooth.

Local agents (and modifying files on your machine) are still sometimes preferred, but I'm excited to make cloud computers easier. You get a secure sandbox + Linux VM you can control, and you can kick off these agents from web/mobile/desktop/Slack/API/more!

Love this so much. Does it only work for web projects? Would love if there is a way to setup Mac and IOS apps with this.

If you're using something like React Native that would work. Think iOS apps are trickier. But you can use this for other desktop software, e.g. we use it to test Cursor itself. Anything in a container!

Congrats to all who worked on this feature, as the computer use is really, really impressive. What you can do now from Cursor cloud agents.

Love it! So now we basically don't even need to read the code or test the output manually. Just see the artifact and approve or reiterate.

I still read the code, but the order has flipped. You start by getting a demo to make sure it actually works, then you can nitpick the code or change the abstraction.

perfectly delivered as always 🐐

Does this only work with monorepos

Nope, works with any repos!

@leerob Coming soon!

@leerob What does that mean tho. Will I be able to configure my bazel build setup? Friendly criticism, because I'm glad you guys are building this. If you don't support serious multi-repo projects first class, more people will probably just build their own Ramp inspect type setups

Nice!

This is so rad Lee. Awesome stuff

run this run on by @vanilagy or just straight ffmpeg?

@vanilagy We built a custom video renderer in Rust! Has nice pans, zooms, etc. Think it uses ffmpeg yeah.

This looks excellent @leerob! Now I am wondering: for a web app, would you have the agent test it at the OS level (as done in the video) or still through Playwright? I assume Playwright can be faster and more precise... but will that advantage hold long term?

sick!

wow!!!

just tried the agent onboarding on our monorepo and holy shit this is a gamechanger, the video demos alone saved my team hours of code review meetings this week

insane stuff, congrats!

Thats HUUUUUUGE

Where is the pricing for this?

It's the same token-based pricing as the rest of Cursor

Thank you 🙏

this is presumably to analyze usage during this initial release phase? and process/negotiate volume based discounts from your vendors to accurately price this? from a GTM perspective I like how this feature upsells people into MAX mode API usage, it is a well architected play.

Does this also work with Expo iOS apps?

okay this is genius

looks like most of that work is just in a browser 👀

Is this enabled by default in the desktop app?

You can kick off cloud agents from the desktop app when you have a git repo + connected to GitHub, yes!

what?! wow this is very very interesting 🤔

👀

Crazy use-case

Very cool

so basically cursor now hires junior devs, pays them in compute credits, and films their screen for your review welcome to the gig economy but make it ai

Apple simulators?

Cursor users keep winning 🤍

The video demo feature is brilliant - saves so much back-and-forth explaining what went wrong. I've wasted hours walking through broken deployments when a 30-second screen recording would have solved it instantly.

This is absolutely crazy.

That's really cool! I will have to check this out!

this is awesome

I don't see the update. Ultra subscriber on MacOS

@drew_carson_ai Did you go to

sorry I'm retarded. yes this worked.

@BHolmesDev problem solved lol