using-android

A guide for controlling a real Android phone or emulator through ADB, the Android Debug Bridge, using device-control tools.

In plain words
What is it for?
Use it to capture screens, tap, swipe, type, install or launch apps, run on-device shell commands, transfer files, and coordinate multiple devices.
Why use it?
It explains the connection between the agent, the computer running the device server, ADB, and the Android device, reducing mistakes when driving the device remotely.

Skill for Claude CodeCodex

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add skills/metahub-tech/agent-fleet/using-android
Any agent
npx skills add metahub-tech/agent-fleet --skill using-android
Clone the repo
git clone --depth 1 https://github.com/metahub-tech/agent-fleet

Made for: Claude Code, Codex.

Per session 60 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 4,032 The whole file, excluding the scripts and references it only reads on demand.
Security scan B 2 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5 $0.00060 $0.04032
Opus 5 $0.00030 $0.02016
Sonnet 5 $0.00012 $0.00806
Haiku 4.5 $0.00006 $0.00403

Measured 2d ago against content hash 5c18b7455e2a, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade B, and why

using-android scanned grade B with 2 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Sends data to an external URLmediumData exfiltration

A POST to an outside endpoint may be telemetry or may be exfiltration; either way the mod talks to somewhere, and you should know where.

curl -X POST --data-binary @bg.png \

Makes network callslowCapability

Not a fault in itself. Listed so you know the mod talks to something, and to what.

curl -X POST --data-binary @bg.png \
platforms/android/skills/using-android/SKILL.md · 244 lines

How it starts

The opening of the file, as written. The whole thing — 244 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Using android-device

Drive one or more Android devices via the android-device MCP server (FastMCP, streamable-http on Tailscale, port 8768). The server runs on a PC host (Windows or macOS), and reaches the phone(s) via ADB (USB or Wireless). 30 tools across 10 categories (+ HTTP POST /upload).

Mental model

[You / Agent]  -- streamable-http -->  [PC Host running android-device MCP]  -- ADB --> [Android Phone]

You drive the SERVER. The server drives ADB. ADB drives the PHONE. You don't touch the phone directly.

Critical patterns

Coordinates are device-screen pixels (no Retina-style scaling)

get_screen_size returns the phone's wm size (e.g. {width: 1080, height: 2340}). take_screenshot returns a PNG sized identically. tap(x, y) uses the same coordinate space. No scaling math needed — phones don't have macOS-style logical/physical pixel split.

get_screen_size                  # {"width": 1080, "height": 2340}
take_screenshot                   # PNG also at 1080x2340
tap(x=540, y=1170)                # center of screen

tap, NOT click

Phones use touchscreens. The tool is tap, not click. Calling click will fail (it's a Windows / macOS thing).

Keyboard: input text vs press_key

What Tool Note
Type into a focused input type_text("hello world") Spaces auto-escaped to %s. Newlines / Chinese / emoji NOT supported by adb input text. For SMS specifically, use the intent-extra workaround in Recipes; for in-app search / share, look for an intent the target app accepts with a string extra.
System buttons press_key("back"), press_key("home"), press_key("recent") Aliases for KEYCODE_BACK / HOME / APP_SWITCH
Volume / power press_key("volume_up"), press_key("power"), press_key("wake") Useful for wakelock testing

App lifecycle

list_packages(filter_substring="weibo", only_user=True)   # find installed user apps
launch_app(target="com.sina.weibo")                        # launcher intent (default)
launch_app(target="com.example", activity=".MainActivity") # explicit activity
current_app()                                              # what's in front
terminate_app(target="com.sina.weibo")                     # force-stop
install_app(path="C:\\\\path\\\\to\\\\foo.apk", replace=True)
uninstall_app(target="com.example.foo")

Read the full file on GitHub · 244 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 2d ago First seen · 244 lines · 60 tokens per session scan B 5c18b7455e2a

Subscribe to this mod's changes

using-android is a skill published in the GitHub repository metahub-tech/agent-fleet (2 stars, last pushed 1mo ago), licensed Apache-2.0. It adds 60 tokens to every session and 4,032 once invoked, about $0.0003 per session on Opus 5. A static security scan graded it B with 2 findings (sends data to an external url, makes network calls). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.

Related

Other skills, from other repositories

imagegen-frontend-mobile

Elite mobile app image-generation skill for creating premium, app-native screen concepts and flows. Designed for iOS, Android, and cross-platform mobile products. Prioritizes clean hierarchy, comfortably readable text, strong multi-screen consistency, controlled color palettes, non-generic creative direction, textured…

Leonxlnx/taste-skill · 126 tokens

canvas

Display, present, and render HTML content on connected Otto nodes (Mac app, iOS, Android). Show on device, preview on mobile, push to screen, or navigate to a URL on any connected node. Use for games, visualizations, dashboards, interactive demos, and live-reloading development previews. Supports presenting, hiding…

elizaOS/eliza · 91 tokens

serve-sim

Control and stream a running iOS, iPad, or Apple Watch Simulator with npx serve-sim. Use for simulator preview, taps, gestures, hardware buttons, rotation, camera injection, permissions, accessibility, and CoreAnimation debug.

EvanBacon/serve-sim · 52 tokens

phone-harness

Control the user's phone — iPhone through the Mac's iPhone Mirroring window, or an Android over adb: open apps, tap, type, swipe, read the screen.

ShawnPana/phone-harness · 40 tokens

app-implement-feature

Guide for implementing features in baguette — a Swift CLI + WebSocket server that drives iOS simulators via private SimulatorKit. Use this skill when: (1) Adding a new gesture, button, keyboard surface, stream format, or device-chrome behaviour (anything that lands across Domain / Infrastructure / App +…

tddworks/baguette · 192 tokens

baguette

Drive iOS simulators programmatically via the baguette CLI — taps, swipes, multi-finger gestures, hardware buttons (Home / Lock / Volume / Action / Power), ASCII keyboard text, and frame capture, all without opening Xcode. Use when: (1) an agent needs to drive a booted iOS simulator from a script — tap a coordinate…

tddworks/baguette · 249 tokens