Give your agents a real phone.

Provision cloud Android & iOS devices, run natural-language tasks on them, stream the live screen, and reach the entire Mobilerun API — without ever holding a Mobilerun key.

naive deployiad · microvm
aappfullstack
ffnserverless
ggpuworkload
build
1.2s
bundle
0.4s
deploy
0.8s
boot
0.3s
ready
live

Paste and ship

Drop this into Cursor, Claude Code, or any agent. It reads the skill manifest and wires up Mobile for you.

Agent prompt

Read https://usenaive.ai/skill.md and set up Naïve for Mobile in my project.

Cloud mobile devices, fully managed.

Naïve owns the Mobilerun credentials and isolates every device to your tenant — provision, drive, and display phones from one bearer token.

Devices

Provision, reboot, reset & stream cloud emulators/devices.

Read docs

Agent Tasks

Run natural-language tasks on a device, with vision & reasoning.

Read docs

Live Stream

Display the device screen with a signed stream URL.

Read docs

Wildcard API

Search & call any Mobilerun endpoint — tap, swipe, apps, proxies.

Read docs

Provision a phone and drive it in a few calls.

The same primitives power your own devices and each of your end-users' devices via forUser(id) — one bearer token.

POST /v1/mobile/devices

Devices

Provision hosted Android/iOS devices on Mobilerun, then reboot, reset, terminate, and get a live stream URL to display the screen. Each tenant only ever sees the devices it provisioned through Naïve.

Provision
Cloud Android & iOS devices.
Live Stream
Signed stream URL + token.
Lifecycle
reboot / reset / terminate.
Per-Tenant Isolation
Scoped to your users.
# Provision a device, wait for ready, then display it
$ naive mobile provision --type dedicated_premium_device --country US

$ naive mobile wait <id>
$ naive mobile stream <id>
POST /v1/mobile/tasks

Agent Tasks

Run a natural-language task on a device — 'Open Settings and enable dark mode' — with optional vision and multi-step reasoning. Poll status, list screenshots, and read the full step-by-step trajectory.

Natural Language
Describe the task in plain English.
Vision & Reasoning
Screenshots + multi-step planning.
Screenshots
Captured at every step.
Trajectory
Replayable action log.
# Run an agent task on a device
$ naive mobile run "Open Settings and enable dark mode" --device <id> --vision

$ naive mobile task <task-id>
$ naive mobile screenshots <task-id>
POST /v1/mobile/call

Wildcard API

Beyond the curated verbs, search a generated catalog of the entire Mobilerun Cloud API and call any operation by id — device tools (tap/swipe/screenshot), apps, proxies, eSIM, flows — with path params substituted and tenant isolation enforced.

Search
Discover any operationId.
Call
Execute by id, any endpoint.
Device Tools
tap / swipe / screenshot / install.
Guarded
Scoped to your own devices.
# Discover, then call any Mobilerun operation
$ naive mobile search "screenshot"

$ naive mobile call take-screenshot --path deviceId=<id>

Put a phone in every agent's hands.

Spin up cloud devices, run tasks, and watch them live — no emulators to host, no Mobilerun key to manage.