A fast, Rust-based headless browser CLI that allows AI agents to interact with web pages through simple, structured commands.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install tekin
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 tekin 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 high-performance, Rust-powered headless browser automation tool designed specifically for AI agents and developers. It bridges the gap between complex browser APIs and simple command-line interactions, enabling seamless page navigation, element interaction, and data extraction. By utilizing a Node.js fallback, it ensures reliability across different environments while maintaining the speed of a native binary.
As one of the most versatile Openclaw Skills, this tool provides a structured way for agents to see and act on the web. It generates interactive snapshots with stable reference IDs, making it incredibly easy to build robust automation scripts or enable autonomous agents to perform tasks without requiring deep knowledge of Playwright or Puppeteer internals.
Agent Browser Use Cases
- Automating complex web workflows and multi-step form submissions programmatically.
- Extracting structured data and accessibility trees from dynamic web applications.
- Performing end-to-end UI testing and visual regression checks via the command line.
- Enabling AI agents to navigate authenticated sessions using saved state files.
- Generating high-quality video recordings and PDF exports of web interactions for debugging and documentation.
How Agent Browser Works
- Initialize the session by opening a target URL using the open command which launches the headless instance.
- Generate a snapshot of the page to identify interactive elements and receive unique, easy-to-use reference tags like @e1 or @e2.
- Execute interaction commands such as click, fill, or scroll using the specific element references provided in the snapshot.
- Re-snapshot the page after any navigation or significant DOM changes to ensure references remain accurate for the current state.
- Save session states or export data like screenshots, HTML, or cookies as needed to complete the automation lifecycle.
Agent Browser Setup
To get started with this skill, ensure you have Node.js and npm installed. Use the following commands to install the CLI globally:
npm install -g agent-browser
agent-browser install
agent-browser install --with-deps
For developers who prefer building from source, you can use the following workflow:
git clone https://github.com/vercel-labs/agent-browser
cd agent-browser
pnpm install
pnpm build
agent-browser install
Agent Browser Data Schema & Taxonomy
This tool manages browser data and interactions using a structured reference system and standard web formats. It is a vital component for those building advanced workflows with Openclaw Skills.
| Data Type | Description | Format |
|---|---|---|
| Page Snapshot | A hierarchical view of the accessibility tree or interactive elements. | Plain Text / JSON |
| Element Refs | Unique identifiers for DOM elements (e.g., @e1, @e2) generated per page load. | String |
| Session State | Persistent cookies, local storage, and authentication tokens for reuse. | JSON |
| Media Outputs | Visual captures of the browser session for audit or debugging. | PNG / PDF / WebM |
| Network Logs | Tracked HTTP requests and intercepted responses for analysis. | JSON |
Agent Browser Advanced Features
- Multi-session management using the session flag to run parallel browser instances with isolated contexts.
- Advanced network routing to intercept, block, or mock API responses during automated sessions.
- Secure state management to save and load authentication sessions, bypassing repetitive login steps for AI agents.
- Full CDP (Chrome DevTools Protocol) support for connecting to existing browser instances on specific ports.
- Comprehensive debugging suite including video recording, trace logging, and headed mode visualization.
- Deep integration with other Openclaw Skills for building autonomous agentic web research and task execution pipelines.
SKILL.md
Loading
Related Openclaw Skills
Protonmail
A powerful integration for AI agents to manage Protonmail accounts using Proton Bridge and the himalaya CLI client.
iCloud Toolkit
A unified CLI integration for managing iCloud Calendar, Email, and Contacts through a single script.
Project Agora
A machine-first API skill for AI agents to discover, participate in, and earn rewards from Project Agora debate forums and job boards.
What Would Elon Do? (WED)
A high-velocity strategic planning tool that converts any business idea into a first-principles execution roadmap.
Auto-Updater
A background utility that automatically keeps your core bot and all installed Openclaw Skills updated to the latest versions daily.
Copywriter
A precision copywriting tool designed to generate user-centric UX microcopy, high-conversion marketing text, and professional product messaging.