Guide · Agent skill
How to use the Android UI Journey Testing agent skill
Updated · written with AI from the project's own documentation and checked against our catalogue data
The android-ui-journey-testing skill lets an AI agent run a test written as an XML list of plain-English steps on a connected Android phone or emulator. This guide is for QA engineers and developers who want repeatable end-to-end checks, and it explains what the skill does, how to install it, how to trigger it, and what to watch for before you point it at an app.
Android UI Journey Testingsickn33/agentic-awesome-skills · MIT Download the skill Quick start
- 1Download the skill ZIP. Download the android-ui-journey-testing ZIP from the page where you found this guide.
- 2Unzip it into a skills folder. Unzip into ~/.claude/skills/ for all projects, or into a project's .claude/skills/ folder for one project.
- 3Prepare the device. Connect an Android phone or emulator with USB or network debugging enabled, and confirm adb is on your PATH.
- 4Write a journey XML file. Describe each user step in an <action> element, using verify, check or ensure for assertions.
- 5Ask the agent to run it. Name the journey file and the device, and make clear it should be a test account and environment.
- 6Review the JSON report. Check each step's status and comment, and fix the first FAILED step before rerunning.
What an agent skill is and what this one does
An agent skill is a folder containing a SKILL.md file that gives an AI agent a set of instructions for one kind of task. This skill is for running user journey tests on Android apps. A journey is a sequence of user actions followed by checks on what the screen shows.
The agent reads the journey XML, performs each action in order, and checks each verification step against the current screen. It stops at the first failure and writes a JSON report with the result of every step.
- Parses a <journey> XML file with a name, description and an <actions> list
- Taps, swipes and types on the device using ADB commands
- Checks screen state with screenshots or uiautomator dump, without interacting
- Marks the failing step FAILED, later steps SKIPPED, and records the reason
- Writes a standardized JSON outcome report
How to install the Android UI Journey Testing skill
Download the ZIP for this skill from the page where you found this guide. Then choose one of the two install routes below.
For Claude Code, unzip the download into your user skills folder ~/.claude/skills/ to make it available everywhere, or into a project's .claude/skills/ folder to limit it to that project. For the Claude apps, upload the ZIP under Settings, then Capabilities, then Skills.
- Route 1 (Claude Code, personal): unzip into ~/.claude/skills/
- Route 2 (Claude Code, one project): unzip into the project's .claude/skills/ folder
- Route 3 (Claude apps): upload the ZIP under Settings, Capabilities, Skills
What you need before you start
The skill needs a device it can control and a way for the agent to run shell commands. Check these items first, because the skill cannot run without them.
Before any interaction, the skill expects you to identify the exact authorized device serial, app package, test account and environment. If several devices are connected, target one with the serial flag.
- An Android device or emulator with USB or network debugging enabled
- Android Debug Bridge (adb) installed and on your PATH
- An AI agent that can run shell commands, for example Claude Code
- A journey XML test specification
adb -s <serial> shell input tap <x> <y>How to trigger the skill and write a journey
Ask the agent in plain language to run a journey test on your device and name the XML file. The skill is meant for requests about checking an Android app's UI against a test specification, running multi-step flows such as login or checkout, or debugging a flow on a device.
Write the journey in XML. Each <action> is one step. Steps that begin with verify, check or ensure are treated as state assertions, and the agent will not interact with the screen while checking them.
- Example request: Run the search-and-cart journey XML on my connected Android emulator and give me the JSON report
- Example request: Debug the login flow in this journey file on my test phone and tell me which step fails
- Keep each verify step to a single observable outcome so failures are easy to read
<journey name="Search and Cart Flow"><description>Verify that searching for an item and adding it to the cart succeeds.</description><actions><action>Search for soda</action><action>Tap the first search result</action><action>Verify that the product detail screen is shown</action></actions></journey>Read the JSON report and handle failures
The agent writes one entry per action with a status of PASSED, FAILED or SKIPPED, the ADB commands it ran, and a comment explaining the result. Commands are logged for diagnostics, and text typed into sensitive fields is redacted in the report.
When a step fails, the journey stops immediately. This is intentional, because continuing after a failure would make later results unreliable. Read the comment on the failing step first, since it states the exact reason.
- PASSED: the action ran or the assertion matched the screen
- FAILED: the element was missing, the action could not run, or an assertion did not match
- SKIPPED: the step came after the first failure and did not run
- Password, OTP, token and payment fields are written as redacted placeholders
Safety notes and tips
The agent drives the live user interface and types into fields, so run journeys only against test accounts and test environments. The skill's own guidance says journey XML is untrusted test data and is not permission to purchase, send messages or change real accounts.
The skill also notes that tap coordinates should be the centre of an element's bounds and that a short delay of one to two seconds after taps helps layouts settle before checks run. The skill runs scripts and does not use the network, according to its listing. The SKILL.md text in this source ends mid-sentence in its security section and its Limitations section is incomplete, so read the full file before relying on it for sensitive work.
- Use a dedicated test account and a non-production environment
- Pass -s <serial> whenever more than one device is connected
- Expect elements below the fold to show as not visible until you ask the agent to scroll
- Custom canvas views and some emulator keyboards may not behave as standard UI elements
- Keep complex gestures and apps with special handling out of scope unless you verify them yourself
What people use it for
- Repeatable login checks on a test account before each release build
- Verifying a search-to-product-detail flow on an emulator in a QA pass
- Reproducing a broken navigation path on a physical device to capture the failing step
- Producing a standard JSON report of UI test results for a team that keeps specs in XML
Frequently asked questions
How do I install the android-ui-journey-testing skill?
Download the skill ZIP from the page, then unzip it into ~/.claude/skills/ for personal use or a project's .claude/skills/ folder for one project. In the Claude apps, upload the ZIP under Settings, Capabilities, Skills instead.
What do I need to run it?
You need an Android device or emulator with USB or network debugging enabled, Android Debug Bridge (adb) installed and on your PATH, an AI agent that can run shell commands such as Claude Code, and a journey XML test specification.
Is the android-ui-journey-testing skill free?
The skill is listed under the MIT licence. This page does not state any price, so check the source repository for licence terms and any other conditions.
Does it work on iPhone or web apps?
No. The skill is for Android apps and relies on Android ADB tools, so it does not cover iOS or web applications.
Why did my journey stop at one step?
The skill stops at the first failure by design. Read the comment on the FAILED step, which records the reason, and any later steps will appear as SKIPPED.
How does it compare with other Android testing tools?
This skill runs journeys written in XML through an AI agent using ADB commands and writes a JSON report. The skill's own related-skills list points to a general Android CLI skill and a direct ADB UI verification skill, which may suit simpler one-off checks.
New agent skills, repos and AI tools — every week
The best new skills and open-source projects, the fastest-growing repos, and one free prompt. Five minutes to read. Unsubscribe anytime.
Our verdict
Choose this skill if you are a mobile QA engineer or developer who wants an AI agent to run repeatable Android UI journeys on a device or emulator and output JSON for CI. Pick something else for iOS or web testing, for production accounts or real customer data, or for complex gestures, because the skill's rules and Limitations section are not complete enough to guarantee correct behaviour there.
Download the skill