Cursor on Android: What Actually Works Today

No. Cursor ships a native mobile app for iPhone and, since July 29, 2026, iPad, in public beta since June 29, 2026, and its documentation states the platform boundary directly: iPhone with iOS 26.0 or later, iPad with iPadOS 26.0 or later, English, Android planned. No date has been published, and as of August 2026 there is no Android beta or waitlist to join.

The question is live enough that Cursor has answered it in public. A feature-request thread on the Cursor forum, opened July 12, 2026, asked for a native Android app to match the iOS beta. A Cursor team member replied that there is no native Android app yet and pointed Android users at the workaround this guide covers, the web surface at cursor.com/agents, with no timeline attached.

More than most people expect, because the workaround is not a stripped-down page. cursor.com/agents is the full web surface for Cursor's cloud agents, it loads in any mobile browser, and it shares one backend with the iOS app and the desktop Agents window. An agent you start from an Android phone is the same agent your desktop sees. The table draws the line between what works today and what stays on iOS.

Plan around the merge row. Reviewing on the web surface and merging on your git host splits one decision across two apps, which reads as a footnote until you are doing it on a train with one bar of signal. If your Android engineers merge often, get them signed in to the git host's own app before anyone needs it.

Sort the rest of that table by where the agent executes.

Local runs, the ones on your own machine, are where the platform gap actually bites, since Remote Control is documented for the iOS app alone. If most of your team's work happens on their own checkouts, the phone is your second problem, and the Remote Control guide explains what the iOS half does.

This is first-party, not a hack: Cursor's web launch post says the agents surface can be installed as a Progressive Web App for a native feel on iOS and Android. On Android the install takes under a minute.

1. 1Open cursor.com/agents in Chrome on your Android phone and sign in to your Cursor account.
2. 2Accept Chrome's install prompt if it appears, or open the browser menu and choose Add to Home screen (labelled Install app on recent Chrome builds).
3. 3Confirm. The Cursor icon lands on your home screen, and the app opens full-screen without browser chrome.
4. 4Run one small agent task end to end from the PWA, so your login and repo connection are warm before you need them away from a desk.

Step four is the one that looks optional, and it is the one I would insist on. Installing and signing in both go fine on a desk connection; what you have not exercised is the part with your repository in it, and that means picking the right project and branch on a small screen and watching a run finish. Two minutes at a desk buys that.

Installed web apps are usually sold on working offline, and that is the part of the pitch that does not carry over here. The agent runs on Cursor's machines and the PWA is only the window you watch it through. Lose signal and you lose the view, though the run itself keeps going on Cursor's side.

Slack is the documented path. Cursor's agents integration posts a notification when a task completes, and it works in the other direction too: mention @Cursor in a Slack conversation to start an agent from your phone without opening the web surface at all. For an Android user, connecting Slack turns a fire-and-forget agent run into something you can actually walk away from.

The common failure here costs you time and nothing else. You start a run from the PWA, lock the phone, nothing arrives, and you look again well after the agent finished. So connect Slack before the first silent run.

Team size changes how I would wire Slack up, mostly because of the phone keyboard. Two or three people can share one channel and put the repository in the sentence they send. Past that, the setup worth doing is a named multi-repo environment, so nobody is choosing repositories one-handed, and Cursor's July 2026 Slack update posts a plan before it starts, which is the cheapest place to catch a task aimed at the wrong project. The integrations guide has the rest.

Check the things that are not the phone. Start with the account you are signed into, then the git connection behind the repository you picked. Cloud agents also require Privacy ModeCursor's setting that guarantees code data is not used for training by Cursor or its model providers, and that an admin can enforce org-wide; data-retention terms are a separate, contractual layer. Press Enter for the full definition. rather than Legacy, and that is an account setting, so the handset has nothing to do with whether it passes.

The cheapest test skips the phone entirely. Open cursor.com/agents in a desktop browser and start the same task there. If it fails in both places the phone was probably never the variable. Reinstalling the PWA is the usual instinct, and it hands you a fresh icon with the same session, so leave that until last.

Cursor has published no date. "Planned" in the iOS documentation is the entire official statement, and the team's forum reply in July 2026 added acknowledgment but no schedule. Any specific window you see elsewhere is a guess.

One structural point runs in Android's favor. All of Cursor's mobile surfaces sit on a single shared backend, so an Android app would be a new client on infrastructure that already works, rather than a second system.

I read that as a timing signal at first, and it is not one. Effort and intent are separate questions, and Cursor has committed to nothing. So treat the PWA plus Slack as this quarter's Android setup, revisit it when the changelog moves, and let the main mobile guide cover what the shared backend can already do.