A high-performance Rust-based headless browser CLI that enables AI agents to navigate, click, and extract data from the web using Openclaw Skills.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install agent-browser-6aigix9qi2tu
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-6aigix9qi2tu 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 automation tool designed to give AI agents a physical presence on the web. Built with a fast Rust core and a Node.js fallback, it allows agents to perform complex actions like clicking, typing, and form filling using a unique reference-based system. By integrating this tool into your Openclaw Skills library, you enable agents to interpret web pages as structured data and interact with DOM elements with high precision.
The tool solves the common problem of AI agents struggling with dynamic web content. It provides a robust interface for navigating the modern web, supporting everything from simple page loads to complex state management across multiple sessions. Whether you are building a research assistant or a testing suite, this skill provides the necessary bridge between LLM reasoning and browser execution.
Agent Browser Use Cases
- Automating multi-step web workflows and complex form submissions.
- Extracting structured data from dynamic, JavaScript-heavy websites for research.
- Performing automated UI and regression testing in headless environments.
- Creating step-by-step video recordings and traces of agent interactions for debugging.
- Managing persistent sessions with saved authentication states to bypass login flows.
How Agent Browser Works
- Initialize the session by navigating to a target URL using the open command.
- Generate a snapshot of the page to identify interactive elements and receive stable references like @e1 or @e2.
- Execute interaction commands such as click, fill, or hover using these assigned element references.
- Monitor the page state by waiting for specific text, URL patterns, or network idle conditions.
- Capture results through screenshots, PDFs, or machine-readable JSON data outputs before closing the browser context.
Agent Browser Setup
This skill requires the OpenClawProvider to be active on your system. To install the core CLI, use the following commands:
npm install -g agent-browser
agent-browser install
agent-browser install --with-deps
Ensure your environment is configured to recognize the Openclaw Skills provider for seamless agent integration.
Agent Browser Data Schema & Taxonomy
The skill organizes its operations through a combination of temporary references and persistent files:
| Component | Type | Description |
|---|---|---|
| Element Refs | String | Temporary identifiers (e.g., @e1) mapping to DOM elements per page load. |
| Session State | JSON | Files like auth.json that store cookies and localStorage for session persistence. |
| Media Assets | File | Outputs including .webm video recordings, .png screenshots, and .pdf documents. |
| Interaction Logs | JSON | Structured output for every command when the --json flag is enabled. |
Agent Browser Advanced Features
- Named sessions for running and managing multiple isolated browser instances in parallel.
- Deep network interception to route, abort, or mock API requests and responses.
- Comprehensive device emulation including custom viewports, geolocations, and color schemes.
- Integration with Chrome DevTools Protocol (CDP) for advanced debugging and connectivity.
- JavaScript execution via the eval command to perform custom page manipulations directly.
SKILL.md
Loading
Related Openclaw Skills
Hevy CLI Skill
A command-line interface skill that enables AI agents to view, create, and manage Hevy fitness app data including workouts, routines, and folders.
gblog
A comprehensive Command Line Interface for managing Google Blogger posts, schedules, and analytics.
Swift Programming
A comprehensive skill enabling AI agents to write, debug, and optimize Swift code for the entire Apple ecosystem.
Pine Editor Skill
A specialized development tool for creating TradingView Pine Script v6 indicators and automated trading strategies.
Agent Browser
A high-performance Rust-based headless browser CLI designed for AI agents to navigate and interact with the web through structured commands.
Agent Browser
A high-performance, Rust-based headless browser CLI designed for AI agents to interact with the web through structured commands.
Featured*
