Appium MCP Server: Setup for Claude, Cursor and Other Apps

• Finds the phone or emulator to test on

  Lists local Android and iOS devices and picks one. If several are connected it asks you to choose. Skip it for remote Appium servers.

  Try: “Which Android devices are connected?”

  Read · select_device

• Boots an iOS simulator ready for testing

  Boots an iOS or tvOS simulator, downloads and caches WebDriverAgent and starts it on a free port. It can also only boot the simulator.

  Try: “Boot the iPhone 16 simulator and get it ready.”

  Runs code · prepare_ios_simulator

• Signs the test runner for a real iPhone

  Lists your provisioning profiles, then downloads, packages and signs WebDriverAgent for a real device. It needs macOS, Xcode 16 or newer and a device in developer mode.

  Try: “Prepare my iPhone for testing.”

  Runs code · appium_prepare_ios_real_device

• Starts, joins and ends test sessions

  Creates, attaches to, lists, selects, detaches from and deletes Appium sessions, locally or on a remote Appium server.

  Try: “Start an Android session on the emulator.”

  Runs code · appium_session_management

• Finds one element on the screen

  Looks up a single button, field or label by a selector so later steps can use it.

  Try: “Find the Login button.”

  Read · appium_find_element

• Finds all matching elements

  Returns every element on screen that matches a selector.

  Try: “List all the buttons on this screen.”

  Read · appium_find_elements

• Gets the focused element

  Returns the element that currently has focus.

  Try: “What field is focused right now?”

  Read · appium_get_active_element

• Reads an element's text

  Returns the visible text of an element.

  Try: “What does the title say?”

  Read · appium_get_text

• Reads an element's attribute

  Returns one attribute of an element, such as whether it is enabled or checked.

  Try: “Is the Submit button enabled?”

  Read · appium_get_element_attribute

• Types text into a field

  Sets or enters text in an input element.

  Try: “Type my email in the email field.”

  Write · appium_set_value

• Taps, swipes, scrolls and pinches

  Performs touch gestures chosen by an action name. The tool handles the differences between Android and iOS.

  Try: “Scroll down to the Settings entry and tap it.”

  Write · appium_gesture

• Drags one item onto another

  Drags an element or point to a target.

  Try: “Drag the card into the Done column.”

  Write · appium_drag_and_drop

• Runs custom multi-touch moves

  Runs raw W3C action sequences for gestures the standard tool cannot do, such as a three-finger swipe.

  Try: “Do a two-finger rotate on the map.”

  Write · appium_perform_actions

• Presses back, home and device buttons

  Presses navigation keys on Android, or physical buttons such as home and volume on iOS and tvOS.

  Try: “Press the Back button.”

  Write · appium_mobile_press_key

• Hides the keyboard or checks it

  Hides the software keyboard or reports whether it is showing.

  Try: “Hide the keyboard.”

  Write · appium_mobile_keyboard

• Reads or sets the clipboard

  Gets or sets the device clipboard as plain text.

  Try: “Put this code in the phone's clipboard.”

  Write · appium_mobile_clipboard

• Dumps the screen as XML

  Returns the current screen's element tree, or saves it to a file.

  Try: “Show the screen structure.”

  Files · appium_get_page_source

• Takes a screenshot

  Captures the screen and saves it to the screenshots folder, which defaults to the system temp folder.

  Try: “Take a screenshot of the current screen.”

  Write · appium_screenshot

• Records the screen

  Starts and stops a screen recording and saves it as an MP4. On iOS it needs ffmpeg.

  Try: “Record the checkout flow.”

  Write · appium_screen_recording

• Gets the screen size

  Returns the width and height of the device window.

  Try: “How big is the screen?”

  Read · appium_get_window_size

• Reads or rotates the screen

  Gets or sets portrait or landscape orientation.

  Try: “Rotate to landscape.”

  Write · appium_orientation

• Handles pop-up alerts

  Accepts, dismisses or reads a system alert on the device.

  Try: “Accept the permission dialog.”

  Write · appium_alert

• Switches between native and web views

  Lists the app's contexts and switches between native and webview contexts.

  Try: “Switch to the webview.”

  Write · appium_context

• Starts, stops, installs and removes apps

  Activates, terminates, installs, uninstalls, backgrounds and checks apps, clears an app's data without uninstalling it, lists apps and opens deep links.

  Try: “Uninstall the test app and install the new build.”

  Delete · appium_app_lifecycle

• Reads or changes app permissions

  Reads and updates Android app permissions or iOS Simulator privacy settings, and resets iOS privacy prompts.

  Try: “Grant the camera permission to the app.”

  Write · appium_mobile_permissions

• Locks, shakes and opens notifications

  Locks or unlocks the screen, shakes the device or opens the notification panel.

  Try: “Lock the phone.”

  Write · appium_mobile_device_control

• Reads device, battery and time

  Returns device information, battery state or the device time.

  Try: “What is the battery level?”

  Read · appium_mobile_device_info

• Reads or fakes the GPS location

  Gets, sets or resets the device's GPS coordinates, on simulators and emulators and on real devices with mock location.

  Try: “Set the location to Berlin.”

  Write · appium_geolocation

• Pushes or pulls a file

  Pushes a file to the device or pulls one from it, using base64 content.

  Try: “Pull the app log from the device.”

  Write · appium_mobile_file

• Reads or changes driver settings

  Reads the current Appium driver settings or updates them, for example idle timeouts.

  Try: “Increase the wait timeout to 10 seconds.”

  Write · appium_driver_settings

• Lists every locator on the screen

  Generates locators for all elements on the current screen. It is heavy, so the docs advise using it only for inspection.

  Try: “Generate locators for this screen.”

  Read · generate_locators

• Writes test code from what you did

  Returns a workflow for the scenario you describe, then generates test code from the interactions the assistant observed.

  Try: “Generate a login test from this flow.”

  Read · appium_generate_tests

Watch out for

• It can uninstall apps and clear app data

  The app tool can remove an app and wipe its data on the connected device. Use a test device and approve those actions. github.com

• It acts on a real device like a user

  It can type text, tap, change permissions, fake GPS and read the clipboard. Do not point it at a personal phone with real accounts. github.com

• Installing from a URL downloads files

  App installs accept web links by default. Set ALLOW_REMOTE_APP_URLS to false to allow local paths only. github.com

• Screenshots and recordings go to disk

  They are saved in the system temp folder unless you set SCREENSHOTS_DIR, so they can show private screens. github.com

• Two optional groups are off by default

  A vision tool turns on with AI_VISION_ENABLED and sends screenshots to the model API you set. A documentation tool group turns on with APPIUM_MCP_DOCS_ENABLED. github.com

Security advisories