Smooth Browser for Openclaw

A specialized browser interface for AI agents to interact with websites, handle authentication, and perform complex web tasks using natural language.

antoniocirclemind
v0.1.0
Feb 5, 2026
0
2.2k
0

Install & Download

1. ClawHub CLI

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

npx clawhub@latest install browser-smooth

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 browser-smooth 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 Smooth Browser?

Smooth Browser is a powerful CLI-driven tool designed to give AI agents a high-fidelity environment for web interaction. It enables agents to navigate complex websites, manage persistent login sessions through profiles, and execute tasks using natural language instructions rather than brittle scripts. By integrating this into your Openclaw Skills library, you allow agents to bridge the gap between static data and the live web, handling everything from form submissions to multi-step research workflows.

The tool is optimized for reliability and persistence, offering features like built-in proxy management and the ability to save browser states (cookies and local storage) across different sessions. This makes it an essential component for developers building agents that require long-term access to gated or dynamic web content.

Smooth Browser Use Cases

  • Persistent authentication for websites requiring logins or 2FA via stored profiles.
  • Automated data extraction from dynamically rendered or JavaScript-heavy websites.
  • Complex multi-step web workflows such as job applications, travel booking, or CRM updates.
  • Human-in-the-loop automation where users can intervene in real-time via live-view links.
  • Testing and QA automation by describing user journeys in natural language.

How Smooth Browser Works

  1. The user initializes a browser session, optionally attaching a persistent profile to maintain state.
  2. Commands are sent to the Smooth CLI using natural language, which the underlying intelligent agent interprets into browser actions.
  3. The agent navigates to the target URL, interacts with DOM elements, and processes the page content.
  4. For data-centric tasks, the user provides a JSON schema to ensure the agent returns structured, machine-readable results.
  5. If manual intervention is needed (e.g., solving a CAPTCHA), the user can access a live view of the session.
  6. The session is closed, ensuring any updated cookies or state changes are saved back to the profile for future use.

Smooth Browser Setup

To get started with this skill, install the library and configure your credentials:

pip install smooth-py
smooth config --api-key <your-api-key>

Verify your installation and configuration:

smooth config --show

You can find this and other Openclaw Skills configurations in your local environment setup.

Smooth Browser Data Schema & Taxonomy

Smooth Browser organizes its operations around sessions and persistent storage entities. Below is how data is structured within the skill:

Component Format Description
Session ID UUID/String Unique identifier generated upon starting a session; required for all subsequent commands.
Profile Directory/ID Stores cookies, local storage, and session state for specific domains.
Response Model JSON Schema Defines the expected structure of data when extracting information from pages.
Metadata JSON Object Arbitrary key-value pairs passed to the agent to assist in form filling or task context.
File ID String References to uploaded documents that the agent can access within a browser session.

Smooth Browser Advanced Features

  • Live View: Generate a temporary URL to view and interact with the agent's browser session in real-time.
  • Structured Data Extraction: A dedicated extract command that bypasses agent steps for faster, schema-based data harvesting.
  • JavaScript Evaluation: Execute custom JS snippets directly in the browser context for advanced DOM manipulation.
  • Multi-Agent Coordination: Support for concurrent sessions, allowing multiple subagents to work on independent web tasks simultaneously.
  • Flexible Device Emulation: Toggle between mobile and desktop user agents to test different site layouts.
  • Proxy Control: Built-in proxying to prevent IP blocking, with the option to disable it for direct connections.

SKILL.md


Loading

Related Openclaw Skills

METADATA

Github Stars: 0
forks: 0

Featured*