Agent Browser Core for Openclaw

A high-performance web automation skill enabling AI agents to interact with the web through structured commands and snapshots.

codedao12
v1.0.1
Feb 6, 2026
22
13.7k
0

Install & Download

1. ClawHub CLI

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

npx clawhub@latest install agent-browser-core

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-core 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 Core?

Agent Browser Core is a specialized Openclaw Skills extension designed for production-ready web automation. It leverages a high-performance Rust-based CLI with a Node.js fallback to provide AI agents with a deterministic framework for navigating websites, capturing snapshots, and executing precise actions. Unlike standard automation tools, this skill focuses on generating compact snapshots with element references and structured JSON output, which are optimized for LLM-driven workflows that require high reliability and execution speed.

Agent Browser Core Use Cases

  • Scrape dynamic web content using AI-friendly element references for precise data extraction.
  • Automate complex multi-step web workflows with deterministic command sequences and guardrails.
  • Capture DOM snapshots to provide context for AI analysis and automated decision-making.
  • Perform headless web testing via CLI with structured JSON feedback for integration testing.

How Agent Browser Core Works

  1. Initialize the agent-browser CLI and confirm the browser runtime environment is properly configured.
  2. Capture a snapshot of the target URL to generate a simplified DOM representation with unique element references.
  3. Identify targets via refs and execute structured commands such as click, type, or navigate.
  4. Validate the success of operations using built-in load-state checks and follow-up snapshots to confirm DOM changes.
  5. Terminate the session or close active tabs to release system resources once the automation task is complete.

Agent Browser Core Setup

To utilize this Openclaw Skills resource, ensure the agent-browser CLI is installed in your environment:

# Install the agent-browser CLI globally
npm install -g agent-browser

# Verify the installation and version
agent-browser --version

Users should also ensure that relevant domain allowlists are configured and that the browser runtime is accessible to the CLI.

Agent Browser Core Data Schema & Taxonomy

The skill organizes its operations through structured JSON outputs and metadata-rich snapshots to ensure consistency across automation tasks.

Component Description
Snapshots Minified DOM representations containing stable element references (refs).
JSON Output Machine-readable results for command execution, errors, and status codes.
Command Map A defined taxonomy of flags and categories for fine-grained browser control.
Safety Profiles Configuration files defining domain allowlists and restricted high-risk actions.

Agent Browser Core Advanced Features

  • Enforced safe mode defaults that automatically block high-risk actions like eval or unauthorized file access.
  • High-speed Rust core with a reliable Node.js fallback to ensure cross-platform compatibility.
  • Deterministic element targeting system that uses stable refs instead of brittle CSS or XPath selectors.
  • Native JSON mode for seamless integration into multi-agent pipelines and automated scripting.
  • Intelligent wait strategies and load-state monitoring designed for modern, asynchronous web applications.

SKILL.md


Loading

Related Openclaw Skills

METADATA

Github Stars: 0
forks: 0

Featured*