For AI agents · iPhone · Android

Hands on the phone.

looktap lets your AI agent see, use and control the apps on your phone. It reads the screen as text and taps by label, on iPhone and Android devices. Then what worked becomes a script or a test your CI runs. No server to start.

npm install -g @looktap/cli

Or without installing: npx @looktap/cli doctor

Read the docs Kickstart your agent

~1 KB: per screen read
0.1 s: per tap
0: dependencies

01 · Look

The screen, as text

ls lists what's on screen, one row per element: a ref, its type, where it is, and its label. Hover a row in the terminal to see which element it is.

01 · Look

agent ➜ iPhone 18 Pro simulator

Every command

43 commands, one line of text each

Read from looktap's own code when this page was built. Each one links to the scene above that shows it. The commands reference has every option.

Benchmark

Against every other agent mobile tool

Nine tools, the same 12 steps in the Settings app on an iPhone simulator and an Android emulator, then on a real iPhone and a real Android phone, 10 runs each, checked by a neutral oracle. What an agent feels is the number of calls, the size of every answer and the setup before the first one.

Agent time on iPhone: 8 steps, plus 2 s per tool call

looktap21 s

Maestro MCP30 s

Appium MCP36 s

mobile-mcp39 s

ios-simulator-mcp44 s

Text sent to the agent for those 8 steps (iPhone)

looktap747 B

Maestro MCP17 KB

mobile-mcp24 KB

Appium MCP102 KB

Setup before the first command (iPhone)

looktap0.5 s

mobile-mcp7.6 s

Maestro MCP18 s

Appium MCP29 s

Agent time on Android: 11 steps, plus 2 s per tool call

looktap28 s

Appium MCP49 s

Maestro MCP60 s

mobile-mcp62 s

Agent time on a real iPhone 14 Pro: 12 steps, plus 2 s per tool call

looktap87 s

mobile-mcp104 s

Appium MCP113 s

Agent time on a real Samsung Galaxy A06: 12 steps, plus 2 s per tool call

looktap38 s

Appium MCP55 s

Maestro MCP66 s

mobile-mcp89 s

The full benchmark, including where other tools are faster, what looktap can't do, and how to reproduce it.

Get started

Two minutes to a phone your agent can use

Installnpm install -g @looktap/cli
Check the toolchainlooktap doctor
Boot a devicelooktap boot "iPhone 16 Pro"
Teach your agentlooktap skill --install
Drivelooktap launch com.apple.Preferences --ls

Step 4 installs the skill for every coding agent on your machine: Claude Code, Codex, Cursor, Gemini CLI, GitHub Copilot, OpenCode and more.

Prefer MCP? Claude Code: claude mcp add looktap -- npx -y @looktap/mcp. Chat apps without a shell (Claude Desktop, Cursor): looktap mcp --config prints a ready-to-paste config.

Node 18+. iOS: macOS with Xcode. Android: adb, with an emulator or a device. looktap boot starts a simulator or emulator, with a window or headless. Requirements · Quick start · For AI agents

Paste into Claude Code, Cursor, Codex, opencode…

Install looktap (https://github.com/lakpriya1s/looktap) so you can drive my app on an iPhone or Android device:

1. Run `npm install -g @looktap/cli`, then `looktap doctor`, and fix anything it reports. If no device is up, boot one: `looktap devices --all` lists them, `looktap boot "<name>"` starts one.
2. Run `looktap skill --install` and read the skill it installs.
3. Find my app's bundle id / package name in this repo and write a looktap.config.json with it at the repo root.
4. Launch the app with `looktap launch --ls` and tell me what screen it's on.

Rules: never type real credentials, OTPs or card numbers. When the app needs a sign-in, stop and let me do it, then continue from there.

Using Claude Code?

Install the plugin, then let it set everything up: it installs the CLI, checks your toolchain and gets a simulator, emulator or iPhone ready.

/plugin install looktap --marketplace lakpriya1s/looktap
/looktap:setup

What the plugin adds