Skip to main content

​
Introduction to PhoneClaw

PhoneClaw is an Android automation app that runs on-device workflows and lets you generate automation logic at runtime using ClawScript, a JavaScript-based scripting language built into the app. Inspired by Claude Bot/Claude Code, PhoneClaw attempts to rebuild the agent loop for Android phones natively to act as your personal assistant with access to all your apps.

​
What makes PhoneClaw different

PhoneClaw runs entirely on your Android device using the Accessibility service. There’s no cloud dependency for execution - your automations run locally with vision-assisted UI targeting powered by Moondream.

Voice-driven generation

Generate automations with natural language commands like “open Twitter and click the blue post button every hour”

No root required

Runs on any Android device with developer mode enabled - no rooting or complex setup needed

Vision-assisted targeting

Uses Moondream vision to find and click UI elements by natural language description, adapting to different layouts

JavaScript-based scripting

Write automations in ClawScript, a familiar JavaScript syntax with helper functions for automation tasks

​
What it can do

  • Automate multi-step app workflows on Android using the Accessibility service
  • Generate scripts at runtime for flexible, adaptive automations
  • Use vision-assisted UI targeting to click controls without hardcoded coordinates
  • Read visible on-screen text and values for branching, validation, and handoffs
  • Schedule automations with cron-like timing for recurring tasks
  • Chain actions across apps (browser, email, media, messaging) inside a single flow
  • Build flows that adapt to different device sizes, layouts, and language settings

​
Key features

​
ClawScript automation engine

ClawScript runs inside PhoneClaw using an embedded JS engine and exposes helper functions for automation, scheduling, and screen understanding. It’s designed for fast iteration: write or generate small scripts at runtime, execute them immediately, and adjust based on UI feedback.

​
Vision-powered element detection

Instead of fragile XPath selectors or hardcoded coordinates, PhoneClaw uses vision models to locate UI elements:
  • magicClicker("Create account") - Finds and taps any button matching the description
  • magicScraper("The 2FA code shown in the SMS notification") - Extracts specific text from the screen

​
Voice command interface

Use natural language to describe what you want to automate. PhoneClaw generates the ClawScript for you, schedules it if needed, and outputs an editable file.

​
Real-world use cases

PhoneClaw has been used to automate:
  • Social media posting - Upload videos to TikTok with songs automatically
  • Account creation - Create Instagram accounts with 2FA from email
  • CAPTCHA solving - Navigate through CAPTCHAs using vision
  • Twitter automation - Post tweets in a Waymo using Android XR
  • Cross-app workflows - Chain together browser, email, and messaging apps
PhoneClaw works best with workflows that involve visual UI interaction. For API-based automation, consider using traditional scripting approaches.

​
How it works

  1. Enable accessibility service - PhoneClaw uses Android’s Accessibility API to interact with UI elements
  2. Provide Moondream auth token - Vision features require a Moondream API token for element detection
  3. Generate or write ClawScript - Use voice commands to generate scripts or write them manually
  4. Execute on-device - Scripts run locally using the embedded JavaScript engine
  5. Schedule recurring tasks - Set cron expressions for automations that should repeat

​
Get started

Ready to automate your Android device?

Quickstart

Get PhoneClaw running in 5 minutes

Installation

Step-by-step installation guide

ClawScript API

Learn the ClawScript helper functions

Example Scripts

See real automation examples

​
Community

Join the PhoneClaw community to share automations, get help, and stay updated: