Injecting Input And State logo

Injecting Input And State

Community
skydoves
injecting-input-and-state

Use this skill to drive an Android device from the host shell — inject taps, swipes, text, key events, and drag-and-drop via `adb shell input`; resize the display with `wm size` / `wm density`; flip hermetic-test settings with `settings put global window_animation_scale 0`; reset app state with `pm clear` vs `am force-stop`; launch Activities with `am start -n pkg/.Activity`; and use the modern `cmd <service>` wrapper (`cmd wifi`, `cmd connectivity`, `cmd uimode night`, `cmd locale set-app-locales`) instead of the deprecated `svc` aliases on API 30+. If the user mentions "input tap", "input swipe", "input text spaces", "KEYCODE_BACK", "wm size for screen-size matrix", "settings put animation_scale", "force-stop vs pm clear", "svc wifi no-op", "cmd wifi set-wifi-enabled", or "am start intent extras", use this skill.

Overview

Publisherskydoves
Repositoryandroid-testing-skills
Skill nameinjecting-input-and-state
Stars
338
Forks
14
Bundled files
Instructions only
LicenseApache-2.0
Links
  • Markdown instructions

    A SKILL.md file the model loads on demand, so it only costs tokens when a request actually matches.

  • Works with any LLM

    AI skills are plain Markdown, not provider-specific code, so this works with GPT, Claude, Gemini, Grok, or a local model.

  • Self-contained

    Everything the model needs lives in the instructions — no extra files to sync.

  • Open source

    Published by skydoves on GitHub. Read the source before you install it.

Installation

Install the Injecting Input And State AI skill in TypingMind to use it with any LLM, or drop it into another agent that reads SKILL.md.

1

Install in TypingMind

TypingMind installs a skill straight from its GitHub folder — it reads SKILL.md, bundles the resource files, and stores the result locally.

  1. Open the app and go to Plugins → Skills.
  2. Choose "Install from GitHub".
  3. Paste the skill folder URL below and confirm.
    https://github.com/skydoves/android-testing-skills/tree/main/adb/control/injecting-input-and-state
  4. Enable the skill in any chat where you want it available.
Plugins → Skills → Add skill → From GitHub URL, then paste the folder URL and press Continue.
2

Install in another agent

Any agent that reads the Agent Skills format can use this skill — copy the folder into that agent's skills directory.

Claude Code — .claude/skills
git clone --depth 1 https://github.com/skydoves/android-testing-skills.git /tmp/android-testing-skills
mkdir -p .claude/skills
cp -r /tmp/android-testing-skills/adb/control/injecting-input-and-state .claude/skills/injecting-input-and-state
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Injecting Input And State in any TypingMind chat and the model takes it from there. Its name and description sit in the system prompt, and the moment a request matches, the model loads the full instructions itself — you never invoke it by hand, and it costs no tokens until it is actually used.

The model loads Injecting Input And State on its own as soon as a request matches it.

Works with any AI model

AI skills are plain Markdown instructions rather than provider-specific code, so Injecting Input And State is not tied to the model it was written for. Install it once in TypingMind and use it with GPT-5, Claude, Gemini, Grok, DeepSeek, Mistral, Llama, or a local model you run yourself — all on your own API keys.

  • Loaded only when it is needed

    The system prompt carries just the name and description. The instructions are fetched on the first matching request, so an idle skill costs nothing.

  • Switch models mid-chat

    Because the skill is instructions rather than code, changing model does not break it — the next model reads the same SKILL.md.

Skill instructions

This is the SKILL.md content the model loads. Read it before installing — a skill is instructions your model will follow.

Injecting Input and State — Drive the Device From the Shell

This skill covers the on-device shell surface a test/QA engineer drives over adb to simulate user gestures, override the screen configuration, flip system settings, reset app state, and launch Activities. Transport (adb devices, USB/TCP) lives in ../../architecture/understanding-adb-architecture/SKILL.md; capturing artefacts after driving the device lives in ../../capture/capturing-screenshots-and-screenrecord/SKILL.md.

When to use this skill

  • The developer needs to script a user journey from CI without writing instrumentation: tap, swipe, type text, press hardware keys.
  • A bug repro requires a specific screen size or density (wm size 411x731, wm density 320) without recreating an emulator AVD.
  • The test suite is flaky because animations are on; the developer wants the canonical hermetic preamble (window_animation_scale = 0 etc.).
  • The developer needs to clear app data between runs and is unsure whether am force-stop is enough (it is not).
  • A script uses adb shell svc wifi disable and silently does nothing on API 30+ — the modern cmd wifi set-wifi-enabled disabled is the fix.
  • The developer needs to launch a deep-link Activity with typed extras (--ez, --ei, --es).

When NOT to use this skill

  • The bytes need to leave the device (screenshots, logs, DB). Use ../../capture/capturing-screenshots-and-screenrecord/SKILL.md, ../../observability/extracting-logs-with-logcat/SKILL.md, ../../transfer/extracting-test-artifacts/SKILL.md.
  • Driving the device from inside an instrumentation test, not from the host. Use ../../../instrumentation/scenarios/launching-activities-with-activityscenario/SKILL.md or UiAutomator.
  • Installing or uninstalling APKs. Use ../../apps/installing-and-managing-apps/SKILL.md.
  • Connecting to the device in the first place. Use ../../devices/connecting-to-devices/SKILL.md or ../../devices/connecting-over-wifi/SKILL.md.

Prerequisites

  • A reachable device or emulator (adb get-state returns device).
  • Platform Tools 23+ on the host. Quoting through adb shell is double-shell (ssh-style) since Platform Tools 23 — load-bearing for any string with spaces (developer.android.com/tools/adb#shellcommands).
  • For cmd wrapper APIs and reliable shell exit codes: device API 24+ (cmd introduced API 24).

input — synthesised user events

input is the on-device wrapper around InputManager.injectInputEvent(). The full grammar:

input [<source>] <command> [<arg>...]

<source> defaults per command (touchscreen for tap/swipe, keyboard for text/keyevent).

Tap, swipe, draganddrop

bash
adb shell input tap <x> <y>
adb shell input swipe <x1> <y1> <x2> <y2> [<duration_ms>]
adb shell input draganddrop <x1> <y1> <x2> <y2> [<duration_ms>]
adb shell input roll <dx> <dy>                        # rotary/trackball axes
adb shell input motionevent <DOWN|UP|MOVE|CANCEL> <x> <y>

swipe defaults to ~300 ms when <duration_ms> is omitted. To synthesise a long-press, swipe in place for the long-press timeout: input swipe <x> <y> <x> <y> 1000. draganddrop differs from swipe in that it emits the drag-and-drop gesture sequence (DOWN → MOVE → UP) consumable by View.OnDragListener, not the same event stream as swipe.

Text input — quoting trap

bash
adb shell input text "<string>"

Caveats (from developer.android.com/tools/adb):

  • Spaces MUST be encoded as %s. input text "hello world" types hello, then nothing.
  • Shell metacharacters (', ", &, (, ), <, >, |, ;, \, backtick) MUST be escaped twice — once for the local shell, once for the device-side shell.
  • Non-ASCII / Unicode is unreliable through input text. Use UiAutomator UiObject2.setText(...) for those.

Key events

bash
adb shell input keyevent [--longpress|--doubletap] <KEYCODE> [<KEYCODE>...]
adb shell input keyevent <NUMERIC_CODE>

Test-relevant KEYCODE_* (KeyEvent constants from frameworks/base; full table in research note A2):

MnemonicNumericNotes
KEYCODE_HOME3Press Home (does not always wake screen).
KEYCODE_BACK4Back button.
KEYCODE_DPAD_UP/DOWN/LEFT/RIGHT/CENTER19-23TV / focus traversal.
KEYCODE_VOLUME_UP/DOWN/MUTE24/25/164
KEYCODE_POWER26Toggles screen off/on.
KEYCODE_TAB61
KEYCODE_ENTER66
KEYCODE_DEL67Backspace.
KEYCODE_MENU82
KEYCODE_APP_SWITCH187Recents button.
KEYCODE_WAKEUP224Wake without unlocking.
KEYCODE_SLEEP223Force sleep.

Wake + dismiss-keyguard for emulator with no lock:

bash
adb shell input keyevent KEYCODE_WAKEUP
adb shell input keyevent 82                     # menu, dismiss the AOSP swipe-up

Source prefixes

input <source> <command> maps to InputDevice.SOURCE_*. Use a source prefix when a screen has multiple input listeners or when targeting non-touch hardware:

SourceTypical use
touchscreenDefault for tap, swipe, draganddrop.
touchpadExternal touchpads (Chromebook / Android tablet keyboard).
dpadTV/focus traversal: input dpad keyevent KEYCODE_DPAD_DOWN.
keyboardDefault for text, keyevent.
mouseMouse buttons + scroll.
trackballTrackball roll.
gamepad / joystickGamepad buttons / joystick axes.
stylusPressure-aware touch (S-Pen).

wm — display size and density overrides

wm is the shell wrapper for IWindowManager. The test-relevant subcommands:

bash
adb shell wm size                               # read current logical size
adb shell wm size 1080x1920                     # override
adb shell wm size reset                         # restore device default

adb shell wm density                            # read current dpi
adb shell wm density 320                        # override
adb shell wm density reset

adb shell wm dismiss-keyguard                   # insecure keyguard only
adb shell wm user-rotation [free|lock [<rotation>]]   # 0,1,2,3 = 0/90/180/270

Use case — screen-size / density matrix without recreating the emulator:

bash
# Per configuration in the matrix
adb shell wm size 411x731
adb shell wm density 240
# … run tests …
adb shell wm size reset
adb shell wm density reset

Caveats:

  • MUST call reset in @AfterClass / suite teardown — otherwise the device stays in the modified configuration after the run, breaking later runs and any human user.
  • wm size / wm density trigger a configuration change that recreates the foreground Activity. Treat them as between-test knobs, not mid-test.
  • am display-size / am display-density are equivalent legacy spellings; wm is preferred.

settings — hermetic test state

bash
adb shell settings [--user <id>] get   <namespace> <key>
adb shell settings [--user <id>] put   <namespace> <key> <value> [<tag>] [default]
adb shell settings [--user <id>] delete <namespace> <key>
adb shell settings [--user <id>] reset <namespace> [<package>] [untrusted_defaults|untrusted_clear|trusted_defaults]
adb shell settings [--user <id>] list  <namespace>

<namespace> is one of global, system, secure.

The hermetic preamble — animation knobs

Set all three to 0 for stable instrumented and screenshot tests:

bash
adb shell settings put global window_animation_scale 0
adb shell settings put global transition_animation_scale 0
adb shell settings put global animator_duration_scale 0

Or via Gradle: testOptions.animationsDisabled = true.

Other test-relevant keys

NamespaceKeyPurpose
systemfont_scaleFloat (e.g. 1.0, 1.3); type-scale tests.
secureaccessibility_enabled1 to enable a11y; pair with enabled_accessibility_services.
secureenabled_accessibility_servicesCSV pkg/.Service for a11y tests.
securedefault_input_methodForce a known IME.
secureshow_ime_with_hard_keyboard1 to keep IME visible on emulators.
securelong_press_timeoutStabilise long-press tests.
globalairplane_mode_onPair with am broadcast android.intent.action.AIRPLANE_MODE.
globalpolicy_controlE.g. immersive.full=* to hide system bars.

App state — am force-stop vs pm clear

CommandKills processes?Clears data/cache/prefs?Resets runtime permissions?
am force-stop <pkg>YesNoNo
pm clear <pkg>Yes (implicitly)YesYes
cmd activity stop-app <pkg>Yes (modern, API 28+)NoNo

For a hermetic between-test reset, use pm clear. For "kill the process so the next launch is cold", am force-stop is enough. See ../../apps/installing-and-managing-apps/SKILL.md for the full pm surface.

Launching Activities — am start

bash
# Most reliable for tests: explicit component.
adb shell am start -n com.example.app/.MainActivity

# Force-stop first, then launch (clean cold start).
adb shell am start -S -W -n com.example.app/.MainActivity

# Launch by action + URI (default handler for https://).
adb shell am start -a android.intent.action.VIEW -d 'https://example.com'

# Pass typed extras.
adb shell am start -n com.example.app/.MainActivity \
  --es screen "settings" --ez debug true --ei count 42

# Return to launcher between tests.
adb shell am start -a android.intent.action.MAIN -c android.intent.category.HOME

Intent specifier flags (abridged from developer.android.com/tools/adb#IntentSpec):

TokenEffect
-a actionIntent action (e.g. android.intent.action.VIEW).
-d data_uriData URI.
-t mime_typeMIME type.
-c categoryCategory.
-n componentExplicit component (pkg/.Activity).
-f flagsHex setFlags() value.
`--es-e key value`
--ez key boolBoolean extra.
--ei key intInt extra.
--el key longLong extra.
--ef key floatFloat extra.
--eu key uriURI extra.
--ecn key componentComponentName extra.
--eia/--ela/--efa key v[,v...]Array extras.
--activity-clear-top etc.Map to FLAG_ACTIVITY_*.
-SForce-stop the target before starting.
-WWait for launch to complete.

cmd <service> — the modern wrapper

cmd is a generic wrapper that calls into a binder service's onShellCommand. Available since API 24. To list services: adb shell cmd -l (or dumpsys -l). Test-relevant entry points:

InvocationEffect
cmd activity stop-app <pkg>Modern equivalent of am force-stop.
cmd activity broadcast …Same as am broadcast.
cmd package install -r -t -g <path>pm install equivalent.
cmd package compile -m speed -f <pkg>Force AOT compilation.
`cmd appops set allowdeny
cmd jobscheduler run -f <pkg> <jobId>Force a JobService to run now.
`cmd uimode night yesno
cmd locale set-app-locales <pkg> --locales en-USPer-app locales (Android 13+).
cmd statusbar expand-notifications / collapseExpand/collapse the shade.
`cmd wifi set-wifi-enabled enableddisabled`
`cmd connectivity airplane-mode enabledisable`
cmd testharness enableReset device to a hermetic test-harness state (Android 10+).

svc deprecation on API 30+

svc data, svc wifi, svc bluetooth are no-ops on API 30+ unless invoked from a privileged shell (corpus §I.10). The preferred replacements are the cmd wrappers above.

Patterns

Pattern: WRONG — svc wifi disable on API 30+

bash
# WRONG
adb shell svc wifi disable
# WRONG because: silently no-ops on API 30+ since the svc CLI was demoted from a privileged
# helper to an ordinary shell command. Exit code 0, Wi-Fi still on. The script proceeds
# under a wrong assumption — exactly the worst kind of failure mode.
bash
# RIGHT
adb shell cmd wifi set-wifi-enabled disabled
adb shell cmd wifi status                       # verify

Pattern: WRONG — am force-stop for a between-test reset

bash
# WRONG
adb shell am force-stop com.example.app
./run-next-test.sh
# WRONG because: am force-stop kills the process but leaves shared prefs, databases, the
# image cache, and granted runtime permissions intact. The "next test" inherits state
# from the previous run.
bash
# RIGHT
adb shell pm clear com.example.app
adb shell pm clear com.example.app.test         # also reset test process state
./run-next-test.sh

(Test Orchestrator with clearPackageData: 'true' is the equivalent inside Gradle — see ../../tests/running-instrumented-tests-via-adb/SKILL.md.)

Pattern: WRONG — input text "hello world"

bash
# WRONG
adb shell input text "hello world"
# WRONG because: spaces in `input text` MUST be encoded as %s. The above types only "hello".
bash
# RIGHT
adb shell input text 'hello%sworld'

Pattern: hermetic preamble + Activity launch

bash
adb shell settings put global window_animation_scale 0
adb shell settings put global transition_animation_scale 0
adb shell settings put global animator_duration_scale 0
adb shell pm clear com.example.app
adb shell am start -S -W -n com.example.app/.MainActivity

-S force-stops first; -W blocks until the launch completes.

Pattern: TV / focus navigation via dpad

bash
adb shell input dpad keyevent KEYCODE_DPAD_DOWN
adb shell input dpad keyevent KEYCODE_DPAD_DOWN
adb shell input dpad keyevent KEYCODE_DPAD_CENTER

Source-prefixing dpad ensures the events route to TV-style focus handlers, not the touchscreen pipeline.

Mandatory rules

  • MUST use cmd wifi set-wifi-enabled / cmd connectivity airplane-mode / cmd bluetooth_manager enable on API 30+. MUST NOT rely on svc wifi / svc bluetooth / svc data — they no-op silently on modern releases.
  • MUST use pm clear <pkg> (not am force-stop) when the goal is a hermetic between-test reset. force-stop only kills processes; data is untouched.
  • MUST encode spaces in input text as %s; MUST double-shell quote (ssh-style) any string with shell metacharacters. Platform Tools 23+ (developer.android.com/tools/adb#shellcommands).
  • MUST call wm size reset and wm density reset in suite teardown when the suite overrode them. A leaked override breaks every subsequent run on that device.
  • MUST set the three animation scales to 0 (or use testOptions.animationsDisabled = true in Gradle) before any UI-asserting test run.
  • MUST NOT use Thread.sleep between adb shell input calls to "let the UI catch up". Inside an instrumentation test, use Compose's idle synchronisation (../../../compose/synchronization/synchronizing-with-idle/SKILL.md); from a host script, use adb shell am start -W to block on Activity launch and adb shell wait-for-device between reboots.
  • MUST NOT assume adb shell am instrument returns a meaningful exit code without -w. See ../../tests/running-instrumented-tests-via-adb/SKILL.md and ../../automation/scripting-adb-for-ci/SKILL.md.
  • PREFERRED: am start -S -W -n pkg/.Activity for cold-start launches in tests — -S resets, -W blocks until ready.

Verification

  • adb shell settings get global window_animation_scale returns 0 (or 0.0) before tests run; same for transition_animation_scale and animator_duration_scale.
  • No script in scripts/ or .github/workflows/ uses svc wifi / svc bluetooth / svc data.
  • Every am force-stop between tests is paired with (or replaced by) pm clear.
  • No input text "<string with spaces>" exists; all spaces are %s.
  • Every wm size <override> in setup has a matching wm size reset in teardown.
  • CI passes 50 consecutive runs without a "device entered prior config" flake.

References

  • developer.android.com/tools/adb — am, pm, input, wm, settings, intent specs (https://developer.android.com/tools/adb).
  • developer.android.com/tools/adb#shellcommands — quoting through adb shell since Platform Tools 23.
  • developer.android.com/reference/android/view/KeyEvent — KEYCODE_* constants table.
  • developer.android.com/training/testing/instrumented-tests/stability — the canonical animation-scale-zero recipe.
  • developer.android.com/studio/test/advanced-test-setup#use-gradle-managed-devices — testOptions.animationsDisabled.
  • Research note tasks/research/A2-adb-shell-commands.md — full input / wm / settings / am / pm / cmd / svc tables and verbatim quotes.
  • Sibling skill: ../../architecture/understanding-adb-architecture/SKILL.md — three-piece architecture, server, daemon.
  • Sibling skill: ../../devices/connecting-to-devices/SKILL.md — adb devices states, wait-for-device.
  • Sibling skill: ../../devices/connecting-over-wifi/SKILL.md — adb pair / adb connect.
  • Sibling skill: ../../apps/installing-and-managing-apps/SKILL.md — full pm install / pm uninstall / pm grant surface.
  • Sibling skill: ../../tests/running-instrumented-tests-via-adb/SKILL.md — am instrument -w -r.
  • Sibling skill: ../../capture/capturing-screenshots-and-screenrecord/SKILL.md — capturing artefacts after driving the device.
  • Sibling skill: ../../observability/extracting-logs-with-logcat/SKILL.md — reading device logs.
  • Sibling skill: ../../transfer/extracting-test-artifacts/SKILL.md — adb pull / adb push / run-as.
  • Sibling skill: ../../automation/scripting-adb-for-ci/SKILL.md — bash idioms, retries, port forwarding.
  • Cross-set: ../../../instrumentation/scenarios/launching-activities-with-activityscenario/SKILL.md — launching from inside an instrumentation test.
  • Cross-set: ../../../fundamentals/strategies/applying-testing-strategies/SKILL.md — when to drive via adb vs instrumentation.

Frequently asked questions

What does the Injecting Input And State AI skill do?

Use this skill to drive an Android device from the host shell — inject taps, swipes, text, key events, and drag-and-drop via `adb shell input`; resize the display with `wm size` / `wm density`; flip hermetic-test settings with `settings put global window_animation_scale 0`; reset app state with `pm clear` vs `am force-stop`; launch Activities with `am start -n pkg/.Activity`; and use the modern `cmd <service>` wrapper (`cmd wifi`, `cmd connectivity`, `cmd uimode night`, `cmd locale set-app-locales`) instead of the deprecated `svc` aliases on API 30+. If the user mentions "input tap", "input...

Why use Injecting Input And State on TypingMind?

Because you install it once and use it with any model. Injecting Input And State is plain Markdown rather than provider-specific code, so the same skill runs on GPT-5, Claude, Gemini, Grok, or a local model — and you can switch model mid-chat without it breaking. TypingMind runs on your own API keys, so you pay providers directly instead of a per-seat subscription, and your skills and chats stay in your own storage.

How do I install Injecting Input And State in TypingMind?

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/skydoves/android-testing-skills/tree/main/adb/control/injecting-input-and-state. TypingMind reads its SKILL.md and installs it as a skill you can enable per chat.

Which AI models can use Injecting Input And State?

Any model you connect in TypingMind. AI skills are plain Markdown instructions rather than provider-specific code, so GPT, Claude, Gemini, Grok, and local models can all load this skill when a request matches it.

How many AI models can I use with Injecting Input And State?

As many as you like. As long as a model supports skills, you can use Injecting Input And State with it — GPT, Claude, Gemini, Grok, DeepSeek, Mistral, Llama and more — all on TypingMind with your own API keys.

Is the Injecting Input And State AI skill free?

Yes. It is published on GitHub by skydoves under the Apache-2.0 license. You only pay your own AI provider for the tokens you use.

What are AI skills?

An AI skill is a reusable instruction bundle that teaches an AI model how to do one specific task. It follows the open Agent Skills format: a SKILL.md file with a name and description, plus any scripts, templates or reference files the model may need. The model reads the instructions only when your request matches the skill, so an installed skill costs nothing until it is used.

How are AI skills different from plugins or MCP servers?

A plugin or MCP server gives a model new tools to call — code that runs somewhere and returns a result. An AI skill gives the model knowledge and process instead: how to approach a task, which steps to follow, what good output looks like. Skills are plain Markdown, so they need no server, no API key and no runtime, and they work with any model.

View all

Set up your own AI workspace now

Get notified about new features and future giveaways by subscribing to our newsletter 👇