A high-performance CLI-based skill for deterministic, AI-friendly web automation using snapshots and structured JSON commands.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install agent-browser-core-1-0-1
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-core-1-0-1 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?
The Agent Browser Core skill provides a production-ready framework for integrating advanced web automation into your AI workflows. Built on a high-performance Rust core with a Node.js fallback, it enables AI agents to navigate the web with precision using compact DOM snapshots and reference-based interactions. By leveraging this Openclaw Skills component, developers can ensure deterministic outcomes and structured data extraction from any website.
This skill is designed for scenarios where reliability and speed are paramount. It minimizes the token overhead for LLMs by translating complex DOM trees into manageable snapshots, allowing agents to act on elements via specific references rather than raw HTML. This makes it an essential tool for building sophisticated web-aware Openclaw Skills.
Agent Browser Core Use Cases
- Automating complex multi-step web workflows with high reliability.
- Generating compact, AI-optimized snapshots of web pages for element identification.
- Performing deterministic UI testing and data scraping within Openclaw Skills environments.
- Executing secure, sandboxed browser actions with strict domain allowlisting and resource management.
How Agent Browser Core Works
- Initialize the agent-browser CLI with the target URL and optional session profile.
- Capture an initial snapshot of the page state to generate element references.
- Analyze the snapshot to identify interactive elements via unique refs.
- Execute structured commands such as click, type, or navigate using the identified references.
- Verify DOM changes by taking a follow-up snapshot and checking load states.
- Close sessions and clean up browser resources upon task completion.
Agent Browser Core Setup
To get started with this skill, ensure you have the agent-browser CLI and a compatible browser runtime installed. Follow these steps to configure your Openclaw Skills environment:
# Install the agent-browser CLI
npm install -g agent-browser
# Verify installation and browser setup
agent-browser --version
# Run a test snapshot to confirm connectivity
agent-browser snapshot --url https://example.com --json
Agent Browser Core Data Schema & Taxonomy
The skill organizes data through structured JSON outputs and reference maps to maintain state across automation steps.
| Data Type | Description | Key Fields |
|---|---|---|
| Snapshot | A JSON representation of the current DOM state. | url, title, refs, snapshot_id |
| Reference Map | Mapping of shortened IDs to specific DOM elements. | ref_id, tag, text_content, attributes |
| Command Result | Execution logs for CLI actions. | success, duration, output, errors |
Agent Browser Core Advanced Features
- Deterministic execution mode for reproducible AI agent behavior.
- Structured JSON output for seamless machine parsing and integration.
- Support for Safe Mode defaults to prevent unauthorized file access or script execution.
- Resource management controls to handle session persistence and cookie mutation.
- High-risk action guardrails to ensure security within Openclaw Skills ecosystems.
SKILL.md
Loading
Related Openclaw Skills
Lotus Wisdom
A meditative problem-solving framework that uses disciplined pauses and Buddhist wisdom domains to navigate complex paradoxes and creative blocks.
OpenClaw Guide
An authoritative lookup tool for accurate documentation, configuration, and source code implementation details for Openclaw Skills users.
Document Spell Check Skill
An automated utility to detect and fix spelling errors in project documentation across multiple file formats.
ai-note
A Git-based, AI-optimized repository for agents to share, access, and contribute to collective technical documentation.
Code
A professional-grade coding workflow for Openclaw Skills that handles planning, implementation, and rigorous verification.
Gemini Deep Research
A powerful AI agent skill that performs systematic web searches and synthesizes multi-source data into comprehensive reports.
Featured*
