A high-performance browser automation CLI built on Playwright for seamless web interaction and data extraction.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install agent-browser-cli
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-browser-cli 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 Browser?
Agent Browser is a specialized command-line interface developed by Vercel to facilitate rapid browser automation. It serves as a core component for Openclaw Skills that require direct web interaction, such as navigating complex single-page applications or performing repetitive online tasks. Unlike standard automation scripts, it provides a flexible environment to handle dynamic content efficiently through the Playwright engine.
By leveraging this tool, developers can execute complex browser workflows directly from the terminal or via automated scripts. It is particularly effective for scenarios where traditional scraping tools fall short, offering a more responsive and programmable way to manage browser sessions within the Openclaw Skills ecosystem.
Agent Browser Use Cases
- Automated daily sign-ins and digital check-ins for reward systems or attendance tracking.
- High-speed bulk form filling and automated data entry into web applications.
- Capturing screenshots and accessibility snapshots of web pages for visual monitoring or technical research.
- Extracting specific web content or accessibility trees for data analysis within Openclaw Skills.
- Navigating multi-step web workflows that require simulated user interactions like clicks and key presses.
How Agent Browser Works
- Initialize a browser session by opening a target URL using the navigate or open command.
- Generate a snapshot of the page to retrieve the accessibility tree and element references for precise targeting.
- Identify specific UI elements using the generated references (e.g., @e14), CSS selectors, or ARIA roles.
- Execute interactions such as clicking buttons, filling input fields, or selecting dropdown options based on the task requirements.
- Capture necessary data, HTML, or screenshots as output for further processing or storage.
- Terminate the browser session using the close command to release system resources.
Agent Browser Setup
To integrate this functionality into your environment, ensure you have Node.js installed, then execute the following commands:
npm install -g agent-browser
agent-browser install
Once installed, the tool becomes available for any Openclaw Skills requiring browser-level automation.
Agent Browser Data Schema & Taxonomy
Agent Browser organizes data through snapshots and metadata capture. It primarily utilizes the following structure to manage web elements and page states:
| Data Type | Description | Format |
|---|---|---|
| Snapshot | The accessibility tree of the current page used for element discovery | Text |
| Element Ref | Unique identifiers for UI elements (e.g., @e1) generated during snapshots | String |
| Visual Output | Full-page or annotated screenshots for debugging or archiving | PNG |
| Page Metadata | Descriptive attributes including page Title, current URL, and raw HTML | Text/HTML |
Agent Browser Advanced Features
- Intelligent element targeting using ARIA roles (e.g., find role button) and accessibility labels for more resilient automation.
- Support for annotated screenshots to visually debug automation steps and element selection.
- Full compatibility with cron jobs and bash scripts for scheduled automation within Openclaw Skills.
- Session persistence capabilities allowing for complex interactions across multiple pages in a single browser instance.
- Low-level keyboard event simulation (Enter, Tab, Escape) to handle complex UI navigation and modal dialogs.
SKILL.md
Loading
Related Openclaw Skills
book-reader
A comprehensive tool enabling AI agents to search, download, and read books from multiple sources with intelligent progress tracking.
Emperor Claw OS
An advanced operating doctrine for Openclaw Skills that enables a centralized manager agent to orchestrate a distributed AI workforce through the Emperor Claw SaaS control plane.
ScholarGraph-FusionGenes
A professional-grade workflow for systematic cancer fusion gene research, automated data extraction, and structured Excel reporting.
Product Comparison Analysis
A comprehensive tool for deep product research and comparative analysis to drive informed purchasing and competitive intelligence.
PR + Commit Workflow
A specialized skill designed to enforce high-signal, human-verified commit and pull request workflows within AI-driven development environments.
OpenWeather
A Python-powered skill for retrieving global weather data, including current conditions and multi-day forecasts for AI agents.
Featured*
