Agent Browser for Openclaw

A high-performance Rust-based headless browser CLI designed for AI agents to interact with web pages via structured commands.

kdegeek
v1.0.0
Feb 23, 2026
0
1.4k
0

Install & Download

1. ClawHub CLI

The fastest way to install a skill directly from the registry.

npx clawhub@latest install agent-browser-0

2. Manual Installation

Copy the skill folder to one of these locations

Global
~/.openclaw/skills/
Workspace
<project>/skills/

Priority: Workspace > Local > Bundled

3. Prompt Installation

Copy this prompt to OpenClaw to install it automatically.

Help me install agent-browser-0 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 robust tool within the Openclaw Skills ecosystem that empowers AI agents to perform complex web automation tasks. Built with Rust for speed and featuring a Node.js fallback, it allows agents to navigate websites, click elements, and extract data with precision.

By providing a structured interface for headless browsers, it bridges the gap between LLM reasoning and real-world web interaction. It is an essential component for developers building autonomous web agents that need to browse the web like a human would, but with the speed and reliability of a programmatic interface.

Agent Browser Use Cases

  • Automating repetitive web interactions and programmatic form submissions.
  • Extracting structured data and accessibility trees for AI-driven analysis.
  • Executing automated UI testing and visual regression checks.
  • Capturing high-quality screenshots, PDFs, and session recordings for documentation.
  • Managing authenticated sessions across multiple parallel browser instances.

How Agent Browser Works

  1. Initialize the browser environment by opening a target URL through the CLI.
  2. Generate a snapshot of the page to identify interactive elements using stable reference IDs like @e1.
  3. Execute interaction commands such as click, fill, or scroll using the specific element references provided in the snapshot.
  4. Re-synchronize the agent's state by taking new snapshots after page transitions or significant DOM updates.
  5. Save session state or export extracted data in JSON format for further processing by the AI agent.

Agent Browser Setup

To get started with this tool from the Openclaw Skills collection, use the following commands:

npm install -g agent-browser
agent-browser install
agent-browser install --with-deps

Alternatively, you can build from source using pnpm to ensure you have the latest features.

Agent Browser Data Schema & Taxonomy

The skill organizes data through element references and session states to provide a structured environment for AI agents:

Component Description
Element Refs Unique identifiers (e.g., @e1, @e2) mapped to interactive DOM elements.
Snapshots Tree-based representations of page accessibility and interactive nodes.
State Files JSON files containing cookies and localStorage for session persistence.
Media Output Formatted WebM recordings, PNG screenshots, and PDF exports.
JSON Mode Machine-readable output available for all primary commands using the --json flag.

Agent Browser Advanced Features

  • Multi-session management for running parallel browser instances with isolated contexts.
  • Advanced network interception to route, block, or mock API responses.
  • Session persistence using state save and load to bypass repetitive login flows.
  • Built-in video recording and trace generation for deep debugging and audit trails.
  • Comprehensive device emulation including viewport size, geolocation, and media scheme overrides.

SKILL.md


Loading

Related Openclaw Skills

METADATA

Github Stars: 0
forks: 0

Featured*