Cursor for iOS Public Beta Is Live: Start Agents, Review Code, and Merge PRs From Your Phone
Cursor has released a native iOS app, letting developers start Cloud Agents from their phone, remotely control local Agents, watch video demos, give task feedback with voice and screenshots, and even review code and merge PRs on mobile.
Introduction
Cursor has now officially brought its AI coding workflow to the iPhone.
On June 29, 2026, Cursor announced that Cursor for iOS has entered public beta. The positioning is not as simple as “write a whole project on your phone.” It turns the phone into a control panel for Cursor Agent: you can send ideas by voice while you are out, start a cloud agent, receive completion notifications, watch the video demo produced by the agent, then review code from your phone, add feedback, and merge directly at the end.
What makes this interesting is that it pushes the originally desktop-oriented agentic coding workflow one step toward “being able to step in at any time.” Cursor is not just letting you chat on your phone. It is letting the phone participate in the full loop across local work, cloud execution, and code review.
The Core Idea Is Not Writing Code on a Phone, but Managing Agents From a Phone
Judging from the official video, the clearest way to use Cursor for iOS is to treat the phone as a remote control for an AI coding agent.
The video demonstrates three scenarios. The first is asking Cursor to test and use software in the cloud, using the same tools as on a MacBook, and finally sending back a working video demo. The user can then read the code on the phone, add feedback, and merge directly when ready.
The second scenario is voice input. While eating, commuting, or even right before falling asleep, the user can send Cursor a long voice note with their ideas and quickly start several agents. Once the agents finish, iOS shows the status through notifications or Live Activity.
The third scenario is UI feedback. When you see something wrong on screen, you can take a screenshot and annotate it directly with markup, such as noting that “the spacing should be a bit tighter,” then let the agent continue from there. This is closer to how design and frontend development actually work than describing UI issues purely in text.
Key Features Listed by Cursor
Cursor’s official blog breaks the iOS app into several main capabilities:
| Feature | What It Means |
|---|---|
| Start Cloud Agents | Send tasks to the cloud even when you are away |
| Remotely control local Agents | Your phone can take over an agent running on your computer |
| Interact with voice, images, and text | Add ideas to a task without sitting at your desk |
| Review code and video demos | Check the results directly on your phone after the agent finishes |
| Merge PRs | The phone is not just for checking status; it can also make the final call |
| Live Activities and notifications | Come back immediately when a task finishes or gets stuck |
The product direction here is very clear: Cursor wants to move agent workflows beyond the constraint that “the person must be sitting in front of the IDE.” Once agents can run in the background, the phone becomes the entry point for task dispatching, review, feedback, and approval.
Composer 2.5 Also Comes to Mobile
Cursor also mentioned that Composer 2.5 is now available in the iOS app. To celebrate the mobile app launch, all mobile requests for Composer 2.5 are 75% off until July 5, 2026.
This discount sends a practical signal: Cursor is not treating the iOS app as just a notification tool. It wants users to actually start and push tasks forward from their phones. If the cost of mobile requests drops, users are more likely to start agents during spare moments with voice, screenshots, and short instructions, instead of waiting until they are back at the computer.
That said, Cursor is also clear that Cursor for iOS is available in public beta for all paid plan users. In other words, this is not a completely free standalone app. It is part of Cursor’s paid workflow.
What This Means for Developer Workflows
What the Cursor iOS app really changes is not the input device, but the rhythm of work.
The old problem with AI coding agents was this: they can run for a long time, but you often have to wait in front of your computer to see when they need more information, when they finish, and when they are ready for review. With the phone filling that gap, the loop can be split into smaller pieces: send a requirement by voice note when it comes to mind, screenshot and annotate a UI issue when you see it, watch the demo after the agent finishes, then check the diff and merge.
This also makes agentic coding feel closer to “asynchronous development.” The human does not have to sit alongside the agent the entire time, but can still step in at the key points: defining requirements, correcting direction, confirming results, and approving the merge.
I think this will be especially useful for a few types of people:
- Developers who already use Cursor Cloud Agent often
- People who frequently think of product or UI changes while on the move
- Team leads who need to review small PRs, demos, or hotfixes
- People who like using voice to quickly describe requirements, then let the agent break down the task
One Thing to Watch: The More Convenient Mobile Gets, the More Important Permission Boundaries Become
Being able to merge PRs from a phone is convenient, but it also means Cursor iOS is not just a notification center. It is a work entry point with real operating permissions.
The things worth watching next are: whether code review is readable enough on mobile, whether the confirmation flow before merge is clear, how permissions are managed when agents switch between local and cloud execution, and whether long voice requests are easy to track after they are turned into tasks. These details will decide whether it is a fun remote control or something that can actually fit into daily development.
Based on this first public beta, Cursor seems to have found a clear direction: the phone is not meant to replace the IDE. It is meant to fill the missing part of the agent workflow, which is the ability for a person to keep giving instructions, checking results, and making decisions after leaving the computer.
Today’s Take
The launch of Cursor for iOS shows that AI coding tools are moving from “desktop IDE” toward “cross-device agent workflow.”
When coding agents can run in the cloud or in the local background, the phone naturally becomes the control layer. That is also why the official video keeps emphasizing that you can start tasks on your phone, receive notifications, watch demos, give feedback with screenshots, and remotely control agents running on your computer.
If Cursor can improve code review, permission confirmation, task tracking, and the notification experience from here, this iOS app could make agentic coding feel more like a real asynchronous workflow, rather than just an AI feature inside an IDE.