> ## Documentation Index > Fetch the complete documentation index at: https://docs.noqa.ai/llms.txt > Use this file to discover all available pages before exploring further. # Device CLI > Let your coding agent run the device and your app for you — it sees the screen and taps by description, with no locators and no coordinates. Your coding agent can take the phone out of your hands. It opens your app on a real device, sees what's on screen, taps and types its way through, and tells you what happened — you stay in your editor. This page is what it can do there. How agents connect to noqa is on [CLI & MCP](/guide/cli-and-mcp); setup and every option live in the [CLI reference](/docs/cli). ## Connect a device ```bash theme={null} noqa devices ios # list iOS devices & simulators (or: noqa devices android) noqa devices connect # connect to one from the list noqa action activate-app --bundle-id com.example.app # bring your app to the foreground ``` One device is active at a time, and every screen read and action goes to it. ## Read the screen ```bash theme={null} noqa screen # cleaned element tree — meaningful elements with relative coordinates (0–1000) noqa screen --full # raw Appium accessibility tree, unfiltered noqa screenshot ./shot.png # save a screenshot for a visual check ``` Your agent reads the screen on **every** step, so how it reads drives the token bill. The three ways differ enormously — measured on the same screen (the Lingua Talks paywall): | Approach | Tokens | vs `noqa screen` | | -------------------------------------- | ------: | ---------------: | | `noqa screen` — cleaned element tree | \~800 | 1× | | Screenshot (vision) | \~1,500 | \~2× | | `noqa screen --full` — raw Appium tree | \~5,600 | \~7× | * **`noqa screen --full`** is the raw accessibility tree straight from Appium — every wrapper, container, and duplicated label. Complete, but noisy and heavy, and it balloons on complex screens. * **A screenshot** is cheap in tokens, but it's pixels only: the model has to infer where elements are and read small text visually, which hurts tap precision and misses anything off-screen. * **`noqa screen`** is our cleaned element tree — the redundant nodes stripped, the meaningful elements kept with their exact bounding boxes. Here that's **\~7× fewer tokens than the raw tree, and lighter than a screenshot** — while still giving precise, structured positions to act on. ## Act by description Working out pixel coordinates from a screenshot is where agents lose their footing: a tap lands 40 pixels off and the run is wrong from there. So the CLI takes the target in plain language and noqa **grounds** it — finds that element on the current screen and places the touch itself. ```bash theme={null} noqa action tap -d "Blue login button at the bottom" # tap noqa action tap -d "Login button" --double # double tap noqa action tap -d "Login button" --duration 2 # long-press, in seconds noqa action swipe -d "photo carousel, swipe left" # swipe noqa action drag -d "card onto the drop zone" # drag noqa action input -d "email field" --text "hi@noqa.ai" # type into a field ``` Describe what a person would look for — position, colour, and the words on it all help ("the X in the top-right corner of the paywall"). Grounding is free; it only needs an account. It works where locators don't, because it never asks the app for an element tree: a Unity button, a WebView, an ad overlay and a native control are all just things on the screen. If grounding puts the touch in the wrong place, fall back to the relative coordinates you just read from `noqa screen` (`0–1000` on both axes): ```bash theme={null} noqa action tap --x 500 --y 320 # tap noqa action swipe --x1 500 --y1 800 --x2 500 --y2 200 # swipe between two points noqa action input --x 500 --y 640 --text "hi@noqa.ai" # tap a field and type ``` ## Control the app and the system ```bash theme={null} noqa action terminate-app --bundle-id com.example.app # force-quit an app noqa action restart-app --bundle-id com.example.app # terminate and relaunch noqa action background-app # send the current app to the background noqa action open-url --url "https://noqa.ai" # open a deep link or URL noqa action alert --action accept # accept a system alert (or --action dismiss) ``` `restart-app` is how your agent gets back to a known state without touching the device by hand. ## The loop Put together, it's **inspect → act → verify**, one step at a time and never an action fired blind: 1. **Inspect** — `noqa screen` to confirm the target is there and the UI is in the expected state. 2. **Act** — one `noqa action …`, by description. 3. **Verify** — `noqa screen` again, or `noqa screenshot` for a visual check. That loop is enough to check a feature you just built. To turn it into saved tests that run on every release, see [Full-cycle testing](/guide/full-cycle-testing). From a build to passing tests Every command and option