Brave Browser Agent for Openclaw

A Chrome DevTools Protocol (CDP) agent that safely attaches to your running Brave Browser to perform resilient automation, dynamic web scraping, and robust SPA click handling.

mayf3
v1.0.0
Jun 5, 2026
0
604
0

Install & Download

1. ClawHub CLI

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

npx clawhub@latest install brave-browser-agent

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 brave-browser-agent 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 Brave Browser Agent?

The Brave Browser Agent is a powerful, non-intrusive automation tool that connects to an active Brave Browser instance on debugging port 9222. Designed as part of the Openclaw Skills suite, it strictly adheres to a read-and-attach philosophy: it never starts, restarts, or terminates the user's browser, preventing disruption to active login sessions, persistent states, and bookmarks. This agent excels at rendering dynamic pages, bypassing bot detection systems with native mouse events, and offering deep browser coordination.

By leveraging the Chrome DevTools Protocol, this skill provides reliable content extraction, DOM interaction, page state monitoring, and visual debugging capabilities. It serves as an essential, zero-risk interface for AI agents to interact with modern websites exactly like a human operator.

Brave Browser Agent Use Cases

  • Advanced Web Scraping & Content Extraction: Read dynamic single-page application (SPA) content where raw HTTP requests fail, using JS evaluation or DOM snapshots.
  • Automated Social Publishing & Form Filling: Safely interact with complex platform editors (e.g., CKEditor, Lexical, React/Vue SPAs) to automate content distribution with built-in framework state sync.
  • Visual Regression Testing: Take full-page or highlighted-element screenshots during workflows to visually verify layouts or capture page states.
  • Interactive Browser Operations: Mimic realistic human behavior (scrolling, delays, real mouse clicks) on anti-bot protected sites without triggering security blocks.

How Brave Browser Agent Works

  1. Attach to Debugging Port: The agent establishes a WebSocket connection over port 9222 using Chrome DevTools Protocol (CDP) to the active Brave Browser instance.
  2. Tab Exploration & Navigation: It queries open tabs or navigates to target pages without modifying existing login cookies, security contexts, or browser configurations.
  3. Resilient Interaction Execution: Emulates precise user behaviors, like native OS-level click events and keystroke triggers, or fires the advanced click-smart fallback routines for JS frameworks.
  4. State Validation & Response: Confirms page readiness (checking load states, network idle, or element presence) before returning JSON-formatted diagnostic logs or visual outputs.

Brave Browser Agent Setup

Prerequisites

Ensure Brave Browser is running with remote debugging enabled. Open your terminal and start Brave using the debugging flag:

# macOS Command
open -a 'Brave Browser' --args --remote-debugging-port=9222

# Linux Command
brave-browser --remote-debugging-port=9222

Connection Verification

To verify that the agent can successfully connect to Brave, run the following diagnostic checks:

# Run quick connection verification
python3 scripts/check_cdp.py --quick

# Run full suite diagnostic (connection, tabs, script health)
python3 scripts/check_cdp.py

Basic Command Usage

Once connected, execute these basic commands to control the browser:

# List all active browser tabs
python3 scripts/cdp_exec.py list

# Open a website in a new tab
python3 scripts/cdp_exec.py open 'https://example.com'

# Capture a screenshot of a specific tab
python3 scripts/cdp_exec.py screenshot <tab_id> output.png

Brave Browser Agent Data Schema & Taxonomy

Script and Operational Metadata

The Brave Browser Agent structure organizes debugging metrics, action pipelines, and integration scripts as follows:

Script Name Purpose Output Format
check_cdp.py Diagnostic script to test CDP accessibility on port 9222 JSON metrics detailing connected state and tab lists
cdp_exec.py Main execution pipeline for standardized CDP tasks JSON payloads mapping DOM coordinates, values, and operation success
framework_click.py Framework-aware click coordinator for React/Vue Structured JSON showing selected click path and editor synchronization state

Command Output Schema

Commands like list and click-smart return structured JSON telemetry for clean integration with downstream AI agents. This data allows agents to track the active document state, analyze interactive buttons, and identify rendering frameworks (such as React, Vue, or Angular) dynamically.

Brave Browser Agent Advanced Features

  • Framework-Aware Click-Smart Engine: Automatically traverses React Fiber, Vue component trees, and virtual DOMs to invoke event handlers when standard clicks fail.
  • Rich Text Editor Sync: Synchronizes internal states of Lexical, CKEditor, Quill, and ProseMirror prior to click triggers to prevent database submission failures.
  • Built-in Anti-Detection: Emulates trustable OS-level input vectors using Input.dispatchMouseEvent and character-by-character human-like keyboard delays.
  • Deep Element Diagnostics: Provides pagination detection, element highlights, and DOM structure analysis directly from Openclaw Skills integrations.

SKILL.md


Loading

Related Openclaw Skills

METADATA

Github Stars: 0
forks: 0

Featured*