This documentation is available as Markdown for AI agents and LLMs. See the full Markdown index or append .md to any documentation URL.

Give a coding agent such as Claude Code, Codex, or Cursor a real Android Emulator or iOS Simulator on EAS so that it can test its own change and return screenshots or recordings.

Expo Simulators is a limited-access preview. It is not included with paid or free plans and is currently available only to select partners. Join the waitlist if you are interested in trying it.

A coding agent can read and change your code, but it cannot see whether the change works unless it can run the app. Expo Simulators gives the agent a remote Android Emulator or iOS Simulator. The agent installs the app, uses it through a controller, and returns screenshots, recordings, or logs as evidence.

Install the Expo Simulators skill

Start with the official Expo Simulators skill. Install Expo Skills, then ask your agent to use the skill. The skill contains the tested steps for the access check, the build choice, the session, the controller, the evidence, and the cleanup.

Use the skill instead of giving the agent individual commands. The simulator:* commands are new, and the skill tells the agent to check the current --help output when a command fails.

Write a prompt that asks for evidence

Describe the result that you want and the evidence that proves it:

What the agent does

With the skill, the agent follows the same loop each time:

1. It runs eas simulator:availability --json. If Expo Simulators is not available for the account, it tells you. Then it uses a local workflow instead, such as a local simulator or emulator.
2. It selects a build that fits the goal. For live edits, it uses a development build with Metro. For a fixed screenshot, it uses a build that matches the current source.
3. It starts a named session and confirms that the status is IN_PROGRESS.
4. It installs and opens the app, then uses agent-device commands such as snapshot -i, press, and fill.
5. It captures screenshots for still states and recordings for motion.
6. It stops the session and resets .env.eas-simulator.

Give the agent access

The agent needs an authenticated EAS CLI and a project linked to EAS:

• On your computer, run eas login.
• In a cloud sandbox, CI job, or background agent, set the EXPO_TOKEN environment variable to an Expo access token. These environments cannot use a browser login.

Confirm the setup:

npx --yes eas-cli@latest makes sure that the agent uses a recent EAS CLI.

Watch the agent work

The start command prints a browser preview URL. Open it in a desktop browser to see the simulator while the agent uses it. In Cursor or VS Code, you can open the URL in the Simple Browser panel.

The session page on expo.dev shows an Open preview button that opens the browser preview in a new tab. Android support is still in development, so some features that work on iOS may not work on Android yet.

To see each controller action as it happens, run this command in another terminal:

To confirm that the agent controls the remote device and not a local simulator, see Confirm that the agent controls the remote device.

Set limits for the agent

A session keeps running until someone stops it, or until it reaches its maximum duration or idle limit. Give the agent limits in the prompt, and add a maximum duration to each session:

• Ask the agent to stop each session that it creates when it finishes or fails.
• Ask the agent to start no more than one session at a time, unless the task needs more.
• Tell the agent to use --max-duration-minutes so that a forgotten session stops on its own.

See Expo Simulators usage, limits, and billing and Session lifetime.

Common questions