A versatile mobile automation skill for interacting with iOS and Android devices, emulators, and simulators.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install agent-device
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install agent-device using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Prefer to download?
Get the raw skill files in a ZIP archive.
What is agent-device?
agent-device is a sophisticated automation tool designed to bridge the gap between AI agents and mobile environments. As part of the Openclaw Skills ecosystem, it enables seamless interaction with iOS and Android targets, allowing for complex UI navigation, data extraction, and state management. Whether you are working with physical devices or virtual simulators, this skill provides the necessary interface to perform actions like tapping, typing, and scrolling with precision.
This skill is particularly valuable for developers and QA engineers who need to integrate mobile testing into their AI-driven workflows. By utilizing snapshots and selector-based targeting, agent-device ensures that interactions remain deterministic and reliable, even as app interfaces evolve. It serves as a critical component in the Openclaw Skills library for any project requiring deep mobile OS integration.
agent-device Use Cases
- Automating end-to-end UI testing for iOS and Android applications.
- Debugging application crashes by capturing logs and reproducing sequences.
- Extracting UI information and metadata for accessibility audits or data collection.
- Replaying interaction scripts to verify UI stability after code changes.
- Managing device states, including app installation, launching, and session tracking.
How agent-device Works
- Target Identification: The skill lists and connects to available iOS simulators or Android emulators using the devices command.
- Session Initialization: The open command launches the specified application or URL on the target device.
- UI Discovery: A snapshot is taken to analyze the current screen hierarchy and identify elements via reference IDs or selectors.
- Interaction Execution: Commands like press, fill, or scroll are sent to the device to simulate user behavior.
- Verification and Diffing: Subsequent snapshots or appstate checks confirm the success of the action or detect UI changes.
- Cleanup: The session is closed, and any captured logs or screenshots are saved for analysis.
agent-device Setup
To begin using this tool from the Openclaw Skills suite, ensure you have the necessary mobile SDKs (like Xcode or Android Studio) installed. It is recommended to use a pinned version for stability.
npx --yes agent-device@latest --help
For iOS development, ensure your environment variables for signing and provisioning are configured if using physical devices. Basic usage involves opening a session:
agent-device open MyApp --platform ios
agent-device Data Schema & Taxonomy
agent-device organizes automation data through snapshots, logs, and script files to maintain consistency across the Openclaw Skills framework.
| Data Type | Format | Description |
|---|---|---|
| Snapshots | JSON | UI hierarchy trees with element reference IDs. |
| Logs | Text/Log | System logs stored in ~/.agent-device. |
| Replay Scripts | .ad | Serialized interaction steps for deterministic replay. |
| Screenshots | PNG | Visual captures of the device screen state. |
| Metadata | JSON | Session info including platform, bundle ID, and device status. |
agent-device Advanced Features
- Interactive snapshot diffing to detect UI drift between app versions.
- Deterministic replay maintenance via the replay -u command for updating selectors.
- Low-level system debugging with integrated log clearing and targeted grep functionality.
- Batch execution mode for running complex interaction sequences from JSON files.
- Support for biometric simulation (FaceID) on iOS simulators.
- Trace recording for detailed post-mortem analysis of automation flows.
SKILL.md
Loading
Related Openclaw Skills
Desearch X (Twitter) Search
A real-time X (Twitter) search and monitoring tool for AI agents to find posts, track users, and analyze social engagement.
Desearch Web Search
A high-performance web search skill that delivers real-time search engine results and structured snippets to AI agents.
Desearch Crawl
A specialized tool for extracting clean text or raw HTML from any webpage to power AI-driven research and automation.
AI Search By Desearch
An AI-powered multi-source search engine that aggregates, filters, and summarizes content from across the web and social platforms.
React Native Logs CLI
A lightweight CLI tool for reading React Native Metro logs via CDP, designed for seamless integration with Openclaw Skills.
Clawland
Clawland enables AI agents to play on-chain odd/even games on the Solana devnet by minting GEM tokens from SOL or USDC.
Featured*
