android-emulator
Skill1 starsChanged 58 days agoMITView on GitHub
Launch, install, drive, and capture Android apps on an emulator or attached device through adb with one explicit serial — installs, deep links, adb reverse networking, uiautomator-based semantic UI driving, screenshots, scoped logcat. Use when a task needs Android runtime verification — launching an APK or dev client, driving Android UI, wiring a local bundler/backend to the emulator, capturing screenshots or logs — "run it on Android", "test on the emulator", "adb". NOT for iOS (ios-simulator) or desktop apps (computer-use).
3 other projects use a version of this.
Adopted by
Most recently taken up by
- callstack/agent-devicetheir own version, first seen Nov 5, 2026
- eigenwallet/coretheir own version, first seen Oct 8, 2026
- callstackincubator/agent-devicetheir own version, first seen Oct 8, 2026
Other versions3 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 3
- callstackincubator/agent-device★ 5.3ksame name
- callstack/agent-device★ 5.3ksame name
- eigenwallet/core★ 324same name
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository27 other files it gives its agents.
- CLAUDE.mdCLAUDE.md
- design-craft/CLAUDE.mdCLAUDE.md
- audit-branchSkill
- brainstormSkill
- create-devtoolSkill
- create-featureSkill
- create-infrastructureSkill
- execute-tasksSkill
- finish-featureSkill
- frontend-testingSkill
- generate-feature-docSkill
- generate-prdSkill
- git-commitSkill
- ios-simulatorSkill
- modify-featureSkill
- pipeline-helpSkill
- plan-implementationSkill
- prd-clarifierSkill
- prd-to-uxSkill
- prototypeSkill
- react-clean-architectureSkill
- receiving-code-reviewSkill
- refactoring-uiSkill
- setup-daher-skillsSkill
- systematic-debuggingSkill
- test-mobile-appSkill
- writing-skillsSkill
The file
We split this skill into the rules it gives an agent
7 rules. 0 are common, 4 are distinctive, and 3 are very specific to this project.
0
Common
Most of the 3 edited versions we read keep these.
4
Distinctive
This author's own choices. Worth a look.
3
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 18
Never stop, erase, or reconfigure an emulator another task owns.
- Very specific: line 24
Check presence: adb -s <serial> shell pm path <package> — presence proves the right variant, not native compatibility.
- Distinctive: line 31
After launch, capture a screenshot before interacting — a foregrounded window alone is not evidence the intended app and screen loaded.
- Distinctive: line 41
Prefer semantic automation the session already exposes (an Android MCP/driver).
- Very specific: line 58
Remove reverse rules this session added: adb -s <serial> reverse --remove tcp:<port>.
- Very specific: line 59
Stop the emulator only if this session booted it: adb -s <serial> emu kill.
- Distinctive: line 60
If the user may iterate, keep the emulator, app, and wiring alive and say so.
“Common” is measured: kept by at least 6 in 10 of the 3 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Android Emulator
Operate Android emulators and devices through adb with an explicit serial and evidence-based verification. For the full mobile verification workflow (platform choice, bundler lifecycle, backend wiring), this skill is driven by /test-mobile-app.
Step 1: Confirm availability
Locate the SDK: adb on PATH or $ANDROID_HOME/platform-tools/adb. List targets: adb devices (running) and emulator -list-avds (bootable). If no SDK, emulator, or device exists, report the missing prerequisite — never claim Android verification without a target.
Step 2: Pin one serial
- Pick ONE serial from
adb devices. Prefer an emulator that is already running; boot one only when needed:emulator -avd <name>as a background process this session then owns. - Pass
-s <serial>on every adb command — with more than one device attached, bareadbfails or hits the wrong target. - Never stop, erase, or reconfigure an emulator another task owns.
Done when: adb -s <serial> shell getprop sys.boot_completed prints 1.
Step 3: Install or reuse, then launch
- Check presence:
adb -s <serial> shell pm path <package>— presence proves the right variant, not native compatibility; reinstall after native changes. - Install:
adb -s <serial> install -r <path.apk>. - Launch:
adb -s <serial> shell am start -W -n <package>/<activity>, or by deep link:adb -s <serial> shell am start -W -a android.intent.action.VIEW -d '<scheme>://<route>' <package>
After launch, capture a screenshot before interacting — a foregrounded window alone is not evidence the intended app and screen loaded.
Step 4: Wire networking
- The emulator reaches the host's loopback at
10.0.2.2, not127.0.0.1. - For a bundler or backend that must appear on-device at
localhost, addadb -s <serial> reverse tcp:<port> tcp:<port>— and record it for removal at cleanup. - Physical devices: no
10.0.2.2; useadb reverseor the host's LAN address with the server bound to0.0.0.0.
Step 5: Drive the UI semantically
- Prefer semantic automation the session already exposes (an Android MCP/driver). Otherwise dump the hierarchy:
adb -s <serial> shell uiautomator dump && adb -s <serial> shell cat /sdcard/window_dump.xml - Target stable
resource-id,content-desc, ortextattributes; compute tap points from that node'sbounds— never from guessed screen positions. - Act with scoped inputs:
input tap <x> <y>,input text '<value>',input keyevent <code>,input swipe .... - Re-dump after every navigation or layout change; stale bounds produce false taps. Prefer a deep link over multi-step blind navigation.
Step 6: Capture evidence and logs
- Screenshot:
adb -s <serial> exec-out screencap -p > <file>.pngat each key state and for the final proof. - Logs, scoped to the app:
adb -s <serial> logcat -d --pid $(adb -s <serial> shell pidof -s <package>)— summarize relevant errors, don't dump unbounded logcat.
Done when: the flow's key states each have a screenshot showing the concrete expected content — real data, not an empty or disconnected shell.
Step 7: Clean up what you own
- Remove reverse rules this session added:
adb -s <serial> reverse --remove tcp:<port>. - Stop the emulator only if this session booted it:
adb -s <serial> emu kill. Leave pre-existing emulators running. - If the user may iterate, keep the emulator, app, and wiring alive and say so.
Troubleshooting
more than one device/emulator→ you dropped-s <serial>; add it everywhere.- App can't reach the bundler → verify
adb reverse --listshows the exact port, then relaunch via the dev-client URL/deep link. - App can't reach the backend → use
10.0.2.2on emulators, never127.0.0.1; confirm the backend port from its own output. - Device shows
offline/unauthorized→ accept the device's USB-debugging prompt;adb reconnectonly as a last resort —adb kill-serverdisrupts every other adb session on the machine. - Taps land on nothing → the dump is stale; re-dump and recompute bounds, or use a deep link.
INSTALL_FAILED_UPDATE_INCOMPATIBLE→ signature mismatch with the installed build; uninstall that package first (only if this test owns it).
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

