Agentic infrastructure for mobile and TV apps.

The loop

Every step is a command your agent can run.

1. 01

   Build

   Cloud builds for iOS, Android and TV: simulator builds to test, signed builds to ship. No Mac needed.

   vibeview buildbuilds logs
2. 02

   Run

   A live phone, tablet or TV, started when the agent needs one.

   dev_startlist-devices
3. 03

   Drive

   Read the screen, tap, type, press the remote, read the app log.

   ui-treetaptypepress +21 more
4. 04

   Test

   Saved tests replay the recording and use AI (beta) where the app changed.

   run_groupvibeview run-suitescreate-test
5. 05

   Ship

   A signed build to TestFlight, Google Play or the Amazon Appstore, followed until testers have it on TestFlight.

   vibeview submittestflight

These are examples: there are 25 device commands, plus commands for builds, tests and store details. Command reference →

The product

Built for agents, readable by the people who ship with them.

Kitlist Build 4 · iOS · Production succeeded

Duration3m 49s

Signingios_distribution · attached

TestFlightavailable to testers

Ship an iOS build without owning a Mac

Cloud builds for phones and TVs, signed with your certificates or ones VibeView generates, then sent to TestFlight, Google Play or the Amazon Appstore.

vibeview.io/app/k3t9…Copy

Anyone with the link · no install · no account

Put the app in someone's hands

A designer, PM or client opens your build in their browser from one link.

1. ✓Open the Lisbon tripReplay
2. ✓Tap Add itemReplay
3. ✓Verify Sunscreen lotion is listedAI

Tests in plain sentences

Describe a test, record it once, or import Maestro flows. AI testing is in beta.

TVs, with a remote

Apple TV, Android TV and Roku. The agent names an element and focus moves to it.

$ vibeview dev

Metro connected · iPhone 17 Pro

Saved App.tsx → reloaded on the device

Live development

Run one command in your React Native app and it opens on a cloud device, hot-reloading as you edit.

Testing

Every run shows what happened.

Add an item to the Lisbon trip

passed

Steps

5 / 5

Duration

21.0 s

Replayed / AI

3 / 2

AI cost

$0.0054

1. ✓ Open the Lisbon trip Replay
2. ✓ Tap Add item Replay
3. ✓ Type "Sunscreen lotion" into Name Replay
4. ✓ Tap Add to Lisbon, long weekend AI
5. ✓ Verify Sunscreen lotion is listed under Toiletries AI

Step 4's path is saved. The next run on an iPhone replays it without AI.

The screen at the end of the run

Describe a test in a sentence, record it, or import Maestro flows. Each run keeps a video and the app log, and a failure caused by the app comes with a bug report ready to paste. How testing works →

Customer story

How MwareTV ships operator apps on VibeView.

“Roku used to be the platform we tested last, because it meant a box on someone's desk. Now our TV apps are tested every night on Roku, Apple TV and Android TV, the phone apps on iPhone and Android, and a screen that changed is flagged before our operators ever see it.”

Sander Kerstens
CTO, MwareTV · CSI Awards 2026, Best TV Everywhere

1. Phones and TVs

   Phone tests run on iPhone and Android, TV tests on Apple TV, Android TV and Roku, and each step's screen is compared with its baseline.

2. Any old version

   Builds stay in VibeView, so a fix is checked against a release from two months ago without rebuilding it.

3. Review by link

   Managers and clients check a change in the running app from a link before it ships, with nothing to install.

4. Roku from anywhere

   The Roku is in the cloud, so developers and agents work on it away from the desk, and record tests on it.

Test runs

2,400+

Cloud builds

140+

Platforms, phone to Roku

5

For agents

Plug in wherever your agent runs.

npm install -g vibeview
vibeview login

# finds the agents on your computer and asks which to set up
vibeview agent-setup

vibeview agent-setup --agent claude-code --mcp

vibeview agent-setup --agent cursor --mcp

vibeview agent-setup --agent codex --mcp
# Codex's sandbox has no network by default: in
# ~/.codex/config.toml, under [sandbox_workspace_write],
# set network_access = true

vibeview agent-setup --agent vscode --mcp

vibeview agent-setup --agent gemini --mcp

vibeview agent-setup --agent windsurf --mcp

# Claude Desktop uses the MCP server, no skill
vibeview agent-setup --agent claude-desktop --mcp

# any MCP client (stdio)
command: npx
args:    ["-y", "vibeview", "mcp"]
# or add the skill to any agent:
npx skills add vibeview/skills

# test a simulator build
vibeview build --cloud --profile preview --json
# BUILD_ID = the app build id from that output
vibeview run-suites --tag smoke --platform ios \
  --build "$BUILD_ID"
# then ship a signed one
vibeview build --cloud --production
vibeview submit --platform ios --latest

POST /api/v1/tests/runs
GET  /api/v1/tests/batches/{id}
# webhooks
run.passed · run.failed
submission.succeeded

Agent control docs →

Start free. Upgrade when you outgrow it.

Black Friday: 30% off your first month, or 25% off your first year on yearly billing, for workspaces that haven't had a paid plan. Ends Dec 1, 23:59 UTC.

Plus applicable tax. Cancel anytime. Limits and overage rates are on the pricing page.