Mobile Flow Node

The Mobile Flow node allows you to automate interactions with native mobile applications (Android and iOS) using Appium. It supports real devices, emulators, and cloud services such as BrowserStack, Sauce Labs, and LambdaTest.

Overview

Prerequisites

Appium

The Mobile Flow node requires an accessible Appium server:

Recommended versions (QANode 0.2.2+)

npm install -g appium@3.2.0
appium driver install --source=npm appium-uiautomator2-driver@7.0.0
appium driver install --source=npm appium-xcuitest-driver@10.25.0

For iOS, XCUITest requires macOS with Xcode configured.

Appium Drivers

General Configuration

Session Mode

• New Session (new): Opens a new Appium session on every execution. Ideal for isolated tests.
• Reuse Session (reuse): Reuses a session already created by another Mobile Flow node in the same flow. Useful for splitting long automations across multiple nodes while keeping the same device connected.

Self Healing

When Self Healing is enabled, Mobile Flow can automatically recover elements in actions such as tap and type when the recorded selectors stop working.

On mobile, the recovery considers signals such as:

• accessibility_id
• visible text
• label, name, value, and hint
• resource-id
• native class/type of the element
• nearby context in the hierarchy captured by Appium

This helps with common changes such as:

• small label adjustments
• uppercase/lowercase differences
• text with or without accents
• selector changes while the same element is still on screen

Mobile self healing also respects the active platform:

• it prioritizes iOS-compatible strategies when the execution runs on iOS
• it prioritizes Android-compatible strategies when the execution runs on Android

When a recovery is applied, it appears in the run details with:

• the affected step
• the recovery confidence
• the mobile selector used
• an Apply to Flow action to promote the recovered selector back to the original flow

Mobile Credential

The credential centralizes connection settings and device capabilities. Configure at Credentials → New Credential → Mobile.

Connection Modes

Device Capabilities

Steps (Mobile Steps)

The Mobile Flow node executes a sequence of configurable steps. Each step represents an action on the device.

When the flow is recorded through MCP, steps appear in the visual editor with the same review pattern as Mobile Inspector, including interaction highlights and useful information for reviewing failures.

Use MCP to record or adjust mobile journeys when an AI client needs to interact with Appium, but review the visual editor to confirm that each recorded target matches the intended action.

Available Actions

Mobile Selector Strategies

Tip: The Mobile Inspector automatically generates selectors when you tap elements, with no need to write XPath manually.

Action Details

tap

Taps an element using selector strategies.

If all selector attempts fail and fallback coordinates exist (generated by the inspector), the tap will be performed at the direct coordinates.

tap-coords

Taps at absolute screen coordinates.

type

Types text into an input field.

For text fields, QANode tries to locate the target in layers:

1. strong identifiers such as resource-id, accessibility id, name, and content-desc;
2. hint, label, and text when they belong to the field itself;
3. contextual relationship between the visible label and the nearby input;
4. recorded coordinates as a visual fallback;
5. generic input class only as a last resort.

clear

Clears the content of an input field.

swipe

Swipes on the screen between two points.

scroll

Scrolls content in a direction.

Modes:

wait

Waits for a condition before proceeding.

Modes:

assert

Verifies a condition in the application. If it fails, the step is marked as failed.

Modes:

Assertion results are available in outputs:

{{ steps["mobile-flow"].outputs.asserts.assertionName }}  →  true or false

extract

Extracts text or an attribute from an element.

Extracted data is available in outputs:

{{ steps["mobile-flow"].outputs.extracts.extractionName }}

double-tap

Double-taps an element using selector strategies.

long-press

Performs a long press on an element.

pinch-zoom

Performs a pinch or zoom gesture on an element.

multi-touch

Performs a multi-touch gesture with two simultaneous fingers.

permission

Accepts or dismisses system alerts, or grants/revokes Android app permissions.

Actions:

grantPermission and revokePermission are Android-only. The permission name follows Android format, for example: android.permission.CAMERA.

push-file

Sends a file from QANode to the device before continuing the automation.

Use this step when the application needs a file to exist on the device before opening a native picker, gallery, document picker, or upload flow.

On Android, common paths live under /sdcard/Download/.... On iOS, QANode tries to send the file to the application container when available and uses an Appium/XCUITest-compatible fallback.

pull-file

Downloads a file from the device, saves it as a QANode artifact, and returns a fileRef.

Prefer Exact path when possible. When the application generates a dynamic filename, use folder/pattern to capture the most recent compatible file.

Images, PDFs, text files, spreadsheets, and videos can be captured as long as Appium can read the file from the device or application container.

reset

Resets the application, clearing its state (equivalent to uninstalling and reinstalling).

back

Presses the physical/virtual Back button (Android — keycode 4).

home

Presses the Home button (Android — keycode 3).

key-event

Sends an Android keycode directly to the device.

Common keycodes:

Evidence (Screenshots)

Each step can have individual evidence settings:

Generated screenshots are available in the Evidence tab of the execution result. Taps and swipes are highlighted with a green visual marker over the screenshot.

Outputs

Downloaded files

When a Download file step runs, the output is available by variable name:

{{ steps["mobile-flow"].outputs.files.report.fileRef }}
{{ steps["mobile-flow"].outputs.files.report.name }}
{{ steps["mobile-flow"].outputs.fileRef }}

Use the fileRef to pass the file to Extract File, File to Base64, HTTP Request, components, or other nodes that accept files.

Complete Example

Android app login

Steps configured in the node:

1. wait → Mode: elementVisible, Selector: id=com.myapp:id/email_field (timeout: 10000ms)
2. tap → Selector: id=com.myapp:id/email_field
3. type → Selector: id=com.myapp:id/email_field, Text: {{ variables.USER_EMAIL }}
4. type → Selector: id=com.myapp:id/password_field, Text: {{ variables.USER_PASSWORD }}, Clear: true
5. tap → Selector: accessibility_id=Login
6. wait → Mode: elementVisible, Selector: id=com.myapp:id/home_title (timeout: 15000ms)
7. assert → Name: loginOk, Mode: elementExists, Selector: id=com.myapp:id/home_title
8. extract → Name: welcomeText, Selector: id=com.myapp:id/welcome_message, Attribute: text

Result after execution:

{
  "sessionId": "my-android-session",
  "extracts": { "welcomeText": "Hello, John!" },
  "asserts": { "loginOk": true }
}

Reusing a Session Across Nodes

Node 1 — Mobile Flow (Login)

• Session ID: android-session
• Steps: login to the app

Node 2 — Mobile Flow (Check Order)

• Session Mode: reuse
• Session ID: android-session
• Steps: navigate to orders, check status

Both nodes share the same device and session state.

Tips

• Use the Mobile Inspector to record steps visually — it generates selectors automatically when you tap elements
• In the Desktop edition, Appium starts automatically when a flow with a mobile node runs — no manual configuration needed
• Prefer accessibility_id as the primary selector — it is more stable than XPath and works on both Android and iOS
• Configure screenshots on critical steps to make failure identification easier
• Use Continue on Fail in monitoring assertions to collect multiple results without interrupting the flow
• For apps that take a long time to start, add a wait step with elementVisible before interacting
• A custom Session ID makes it easy to reuse sessions across multiple Mobile Flow nodes in the same flow