Stop queueing iOS reviews behind one Mac.

Give each cloud coding agent a disposable hosted iOS Simulator through remote MCP. Build from Linux, launch the app, inspect the UI, act, and collect logs without handing the agent a Mac or Xcode.

No card · 2 Simulator hours · 1 source build

mobile / pull requests3 Simulators active
Agent queuedemo data
#1842fix checkout address running
#1843onboarding deep link running
#1844profile avatar crop running
#1845receipt accessibility scheduled
Hosted iOS runtimeScale capacity 3 / 10
sim-01
Build succeeded#1842 · launching app
ready
sim-02
Agent acting#1843 · tap “Continue”
live
sim-03
Evidence retained#1844 · screen + logs
recording
OAuth MCP 90 days disposable

Review the run, not a claim that it passed.

Every retained run keeps its screens, actions, logs, and recording together. These are actual AgentCloud product screens with demo data.

AgentCloud run detail showing an iPhone Simulator beside the agent action timeline and retained evidence.
Inspect one change in contextMatch each agent action to the screen and run evidence it produced.Open full resolution
AgentCloud Runs screen listing active, completed, and failed Simulator sessions.
Scan every runStatus, device, duration, events, and retained files stay visible.Open full resolution
AgentCloud Connect screen showing the team MCP URL and OAuth connection steps.
Connect each agentUse one remote URL and authorize team access in the browser.Open full resolution

Replace the local Mac queue with parallel evidence.

Review constraint Local Mac queue AgentCloud
Cloud agent accessCoordinate a local machineRemote OAuth-enabled MCP
Parallel workOne Simulator becomes a queue1, 4, or 10 concurrent sessions by plan
Build sourcePrepare on the MacUpload build-ready source from Linux
Reviewer evidenceScreenshots and logs are scatteredScreens, actions, logs, and recording stay joined
Environment boundaryMac desktop and Xcode are exposedFocused iOS tools; no raw Mac shell

From agent change to reviewable iOS run.

AgentCloud schedules the build and Simulator separately, then keeps the complete run as one review surface.

  1. Upload the iOS target

    The agent sends build-ready source or a Simulator-compatible .app.

  2. Schedule build and session

    A managed Xcode build creates a reusable Simulator artifact when source is provided.

  3. Launch, observe, and act

    The agent uses focused tools to inspect the UI, interact, and read application output.

  4. Review retained evidence

    Your team inspects the screens, actions, logs, artifacts, and full-run recording.

Ten iOS sessions. One remote endpoint.

Scale gives mobile teams and agent products about 841 hours of iOS Simulator usage credit, 10 concurrent sessions, unlimited source builds, and 90-day run evidence for a $999 monthly minimum.

10
concurrent iOS Simulator sessions
~841 hr
usage credit at the Scale active-second rate
Unlimited
source builds after the free test
90 days
of retained run evidence

The exact technical boundary.

A focused hosted iOS Simulator, not a remote Mac desktop.

What can an agent upload?

AgentCloud accepts build-ready iOS source or a Simulator-compatible .app. Device-signed .ipa files are not supported.

Does AgentCloud provide a Mac or Xcode UI?

No. Agents receive focused Simulator tools through MCP. AgentCloud does not expose a Mac desktop, raw worker shell, or Xcode interface.

Will managed source builds work for every project?

The source must include its Xcode workspace or project and the dependencies required to produce a Simulator build. Expo and React Native projects must arrive in that build-ready iOS form.

How do cloud agents connect?

Use the team’s Streamable HTTP MCP URL from an OAuth-capable client. OAuth 2.1 with PKCE handles identity and team authorization in the browser.

What about Android?

Android Simulator support is coming soon. Today, AgentCloud provides hosted iOS Simulator sessions.

Let the next iOS PR run in parallel.

Test one real build with 2 free Simulator hours and 1 source build total.

Start free