Automates Android device interactions for app testing, UI navigation, screenshots, and data extraction. Use when the user needs to control Android devices, inspect app UI, test Android apps, or extract device info.
Automates Android devices for app testing, UI navigation, and data extraction via ADB.
/plugin marketplace add minibear2021/agent-android/plugin install minibear2021-agent-android-plugin@minibear2021/agent-androidThis skill inherits all available tools. When active, it can use any tool Claude has access to.
agent-android connect 192.168.1.5 # Connect to device (optional if USB)
agent-android snapshot # Get elements with refs
agent-android tap :e1 # Tap element by ref
agent-android input "hello" # Type text
agent-android back # Press back button
agent-android connect <host> or ensure USB device is listed in agent-android devicesagent-android snapshot (returns elements with refs like e1, e2)To conserve tokens and improve performance:
snapshot over screenshot: The snapshot command returns a structured text representation of the UI, which consumes significantly fewer tokens than processing raw images. Only use screenshot when visual verification is strictly necessary (e.g., checking layout rendering or image content).agent-android devices # List connected devices
agent-android connect <host> # Connect to TCP/IP device
agent-android push <local> <remote> # Push file to device
agent-android pull <remote> [local] # Pull file from device
agent-android disconnect # Disconnect from all or specific device
agent-android info # Get device info
agent-android shell <cmd> # Run shell command
agent-android exec <args> # Run raw adb command (e.g. exec reboot)
agent-android snapshot # Get UI hierarchy (compact mode by default)
agent-android snapshot -i # Interactive elements only
agent-android snapshot -f # Full output (include resource-ids and structural elements)
agent-android snapshot -d 3 # Limit recursion depth to 3
agent-android snapshot --selector "Login" # Filter by text content (subtree)
agent-android snapshot --selector "role=list" # Filter by role (e.g., list, button, textbox)
agent-android snapshot --selector "id=header" # Filter by resource-id
agent-android tap :e1 # Tap element
agent-android tap 500 1000 # Tap coordinates
agent-android input "hello" # Type text
agent-android key HOME # Press Home button
agent-android back # Press Back button
agent-android scroll down # Scroll down
agent-android scroll left # Scroll left
agent-android scroll 100 500 100 100 # Swipe manually
agent-android wait 2000 # Wait milliseconds
agent-android find text "Login" # Find element info (no click)
agent-android find "text=Login" # Same as above (key=value syntax)
agent-android find text "Login" click # Find element by text and click
agent-android find "text=Login" click # Find element by text and click
agent-android find :e1 click # Find by ref and click
agent-android find role button tap # Find first button and tap
agent-android is visible text "Submit" # Check visibility
agent-android is visible :e1 # Check visibility by ref
agent-android get text resource-id com.app:id/title # Get text content
agent-android get bounds text "Login" # Get element bounds
agent-android get text :e1 # Get text by ref
agent-android check text "I agree" # Check checkbox
agent-android uncheck :e1 # Uncheck element
agent-android select "Option A" # Select option (click text)
agent-android select :e1 "Option A" # Click dropdown :e1 then click "Option A"
agent-android install app.apk # Install APK
agent-android uninstall com.example.app # Uninstall package
agent-android list-packages # List installed packages
agent-android start com.example.app # Start application
agent-android stop com..example.app # Force stop package
agent-android screenshot # Screenshot to screenshot.png
agent-android screenshot out.png # Save to specific file
agent-android record video.mp4 10 # Record 10s video
agent-android start com.example.app
agent-android snapshot
# Output shows: textbox "Email" [ref=e1], textbox "Password" [ref=e2], button "Login" [ref=e3]
agent-android input :e1 "user@example.com"
agent-android input :e2 "secret"
agent-android tap :e3
agent-android wait 2000
agent-android snapshot # Check result
Add --json for machine-readable output:
agent-android snapshot --json
agent-android devices --json
Target specific device by serial:
agent-android -s serial123 tap :e1
agent-android -s 192.168.1.5:5555 snapshot
Use when working with Payload CMS projects (payload.config.ts, collections, fields, hooks, access control, Payload API). Use when debugging validation errors, security issues, relationship queries, transactions, or hook behavior.
Creating algorithmic art using p5.js with seeded randomness and interactive parameter exploration. Use this when users request creating art using code, generative art, algorithmic art, flow fields, or particle systems. Create original algorithmic art rather than copying existing artists' work to avoid copyright violations.