A high-performance automation layer for AI agents to control iOS simulators through both mobile Safari and native applications.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install atl-mobile
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 atl-mobile 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 atl-browser?
atl-browser (Agent Touch Layer) serves as the critical bridge between AI agents and the iOS ecosystem. It provides HTTP-based automation for the iOS Simulator, supporting both web-browser tasks in mobile Safari and deep interaction with native iOS applications. By exposing two specialized servers, it allows developers to navigate, click, and manage mobile environments with the same ease as desktop browser automation frameworks.
This skill is uniquely optimized for Openclaw Skills to enable vision-free automation. It utilizes a sophisticated marking system to identify interactive elements, providing pixel-perfect spatial understanding without the high token costs or latency associated with traditional vision models. Whether you are automating a complex e-commerce checkout flow or inspecting a native app accessibility tree, atl-browser delivers deterministic and reliable results.
atl-browser Use Cases
- Automating complex e-commerce checkout flows within mobile Safari.
- Testing native iOS application functionality and accessibility hierarchies.
- Capturing high-quality screenshots and PDF vision layers for UI state analysis.
- Executing cross-app workflows, such as transferring data between Safari and the Notes app.
- Reducing AI token consumption by utilizing coordinate-based interaction instead of constant vision API calls.
How atl-browser Works
- The skill initializes two distinct HTTP servers on ports 9222 (Browser) and 9223 (Native) within the booted iOS Simulator.
- For web-based tasks, the Browser server implements a marking system that labels every interactive element with a unique numerical ID.
- For native application tasks, the Native server leverages XCTest APIs to expose the accessibility tree as a machine-readable JSON snapshot.
- AI agents or developers send JSON commands via CURL to navigate, tap specific coordinates, or inject JavaScript for direct DOM manipulation.
- The system follows an escalation ladder, moving from fast coordinate-based taps to vision fallbacks and finally JS injection to resolve UI blockers.
atl-browser Setup
To get started with this skill, ensure you have macOS with Xcode installed and then follow these steps:
# Clone the ATL repository
git clone https://github.com/JordanCoin/Atl ~/Atl
# Execute the automated setup script
~/.openclaw/skills/atl-browser/scripts/setup.sh
# Find a suitable simulator UDID (iOS 17+)
xcrun simctl list devices available
# Build and install the ATL app to your target simulator
cd ~/Atl/core/AtlBrowser
xcodebuild -workspace AtlBrowser.xcworkspace -scheme AtlBrowser -destination 'id=<SIM_UDID>' -derivedDataPath /tmp/atl-dd build
xcrun simctl install <SIM_UDID> /tmp/atl-dd/Build/Products/Debug-iphonesimulator/AtlBrowser.app
# Launch the server within the simulator
xcrun simctl launch <SIM_UDID> com.atl.browser
atl-browser Data Schema & Taxonomy
The skill organizes its automation data using the following structures:
| Data Type | Description | Primary Fields |
|---|---|---|
| Accessibility Snapshot | JSON tree of native UI elements | ref, label, type, x, y, isHittable |
| Browser Mark Info | Metadata for tagged web elements | label, tag, text, x, y, width, height |
| Native App State | Status of the currently active app | bundleId, mode, state |
| Vision PDF | Full page PDF with machine-readable text | path, status, id |
atl-browser Advanced Features
- Vision-free spatial understanding via a robust coordinate marking system.
- Seamless switching between Browser and Native modes for cross-application automation.
- Direct JavaScript evaluation for complex web interactions and data extraction.
- Advanced touch gesture support including multi-directional swipes, pinches, and long-presses.
- Native accessibility tree traversal to interact with any standard iOS application via bundle ID.
SKILL.md
Loading
Related Openclaw Skills
trainedby-mcp Personal AI Coach
A personal AI coaching skill that connects health data, fitness goals, and training reflections to your AI agent via the trainedby.ai MCP server.
OpenClaw Starter Guide
A strategic roadmap for deploying, configuring, and optimizing cost-effective multi-agent AI systems from zero to professional scale.
Callmac: Remote Voice Control for Mac
Control your Mac's voice remotely from mobile devices to play announcements, alarms, and stories using high-quality neural TTS.
SiliconFlow Provider Configuration
A comprehensive configuration skill to integrate SiliconFlow's inference platform into OpenClaw for access to over 98 AI models including DeepSeek, Qwen, and Kimi.
notioncli
A high-performance CLI for managing Notion workspaces, databases, and pages without needing manual UUIDs.
Google Search
A powerful search integration allowing AI agents to perform real-time Google searches to retrieve live information from across the web.
Featured*
