A high-performance Rust-based headless browser CLI designed to give AI agents full control over web navigation and interaction.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install agentbrowser
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 agentbrowser 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 fast, Rust-powered automation tool specifically engineered for AI agents to navigate the web. By providing a structured CLI interface with a Node.js fallback, it allows agents to perform complex tasks like clicking, typing, and data extraction through simple commands. As a versatile addition to the Openclaw Skills library, this tool bridges the gap between LLMs and the live web, offering a reliable way to automate browser workflows without the overhead of heavy testing frameworks.
The tool utilizes a unique reference-based system where interactive elements are assigned stable tags (like @e1), making it significantly easier for AI agents to interact with specific UI components accurately. Whether you are building research agents or automation scripts, this skill provides the necessary primitives for deep web integration.
Agent Browser Use Cases
- Automating complex web interactions and multi-step workflows
- Extracting structured data from dynamic or JavaScript-heavy pages
- Programmatic form filling and automated submission for lead generation
- Continuous web UI testing and quality assurance in CI/CD pipelines
- Generating accessibility trees and snapshots for LLM processing
How Agent Browser Works
- Initialize the session by navigating to a target URL using the open command.
- Generate a snapshot of the page to create an accessibility tree with interactive element references.
- Identify specific element tags (e.g., @e1) from the snapshot output to target interactions.
- Execute commands such as click, fill, or scroll using the identified references.
- Re-snapshot the page after navigation or DOM changes to maintain an accurate state for the AI agent.
- Optionally save the session state or export data as JSON, PNG, or PDF.
Agent Browser Setup
To get started with this tool from the Openclaw Skills collection, install the CLI globally using npm and initialize the browser dependencies:
npm install -g agent-browser
agent-browser install
agent-browser install --with-deps
For developers preferring to build from source:
git clone https://github.com/vercel-labs/agent-browser
cd agent-browser
pnpm install
pnpm build
agent-browser install
Agent Browser Data Schema & Taxonomy
Agent Browser organizes interaction data and page metadata through structured outputs, which is ideal for integration into Openclaw Skills workflows.
| Data Type | Format | Description |
|---|---|---|
| Snapshots | Text/JSON | An accessibility tree containing interactive elements with @ref tags |
| Session State | JSON | Portable authentication data including cookies and localStorage |
| Visual Exports | PNG/PDF/WebM | Screenshots, page exports, and video recordings of interactions |
| Network Data | JSON | Detailed logs of intercepted requests and responses |
Agent Browser Advanced Features
- Isolated session management allowing for parallel browser instances using the --session flag
- Full session video recording (WebM) and trace generation for visual debugging
- Advanced network interception to block, route, or mock API responses
- Semantic locators for finding elements by role, text, or label instead of CSS selectors
- State persistence to bypass login walls by saving and loading authentication cookies
- Built-in Chrome DevTools Protocol (CDP) support for connecting to existing browser instances
SKILL.md
Loading
Related Openclaw Skills
Pipeworx Dad Jokes
A dedicated skill for fetching random and searchable dad jokes from icanhazdadjoke.com.
Pipeworx Crypto
An essential skill for accessing live cryptocurrency market data and performing fiat currency conversions directly within your AI agent.
Crossref Academic Papers
A powerful academic research tool to search 150 million scholarly works, retrieve DOI metadata, and analyze journal information.
Unofficial Urban Dictionary API Skill
A specialized skill for AI agents to query slang, urban phrases, and cultural definitions via the Unofficial Urban Dictionary API.
Pipeworx Color API
A professional-grade color identification and palette generation tool designed for seamless integration with AI coding workflows.
Pipeworx Countries
A high-performance MCP skill for accessing global country data including demographics, geography, and languages via the REST Countries API.
Featured*
