A local browser automation powerhouse designed for persistent sessions, AI agent orchestration, and advanced data extraction within the OpenClaw environment.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install browser-use-local
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 browser-use-local 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 browser-use-local?
browser-use-local is a specialized skill that enables high-performance browser automation using the browser-use CLI and Python framework. Unlike standard browser tools, this skill is optimized for local execution within a container or host, offering superior reliability for complex tasks. It is a vital component for developers building Openclaw Skills that require persistent sessions to maintain state across multi-step workflows.
The skill excels in environments where traditional scraping might fail. By providing low-level access to HTML, screenshots, and DOM evaluation, it ensures that your automation logic remains resilient even on heavy, JavaScript-driven websites. It also bridges the gap between web navigation and generative AI by supporting OpenAI-compatible LLMs like Moonshot and Kimi for autonomous agent-led browsing.
browser-use-local Use Cases
- Managing multi-step authentication and navigation flows using persistent browser sessions.
- Running autonomous agents for web research with specialized LLM providers like Moonshot/Kimi.
- Extracting authentication QR codes from login pages using advanced image cropping and base64 decoding.
- Debugging unresponsive web pages by capturing screenshots and raw HTML when standard DOM queries fail.
- Automating data extraction tasks that require custom JavaScript evaluation and link discovery.
How browser-use-local Works
- Initialize a named session to persist cookies and browser state across multiple execution steps.
- Execute browser commands such as opening URLs, clicking buttons, or typing text through the CLI or Python scripts.
- Capture the page state using high-resolution screenshots or structured JSON data for analysis.
- Apply specialized processing scripts to extract visual elements like QR codes or to filter links using regex.
- Integrate with an LLM agent to interpret the page content and make autonomous decisions based on the current browser state.
browser-use-local Setup
To integrate this capability into your collection of Openclaw Skills, follow these installation and configuration steps:
- Activate the dedicated virtual environment:
source /home/node/.openclaw/workspace/.venv-browser-use/bin/activate
- Configure your environment variables for agent-based workflows:
export OPENAI_API_KEY=your_api_key
export OPENAI_BASE_URL=https://api.moonshot.cn/v1
- Verify the installation by launching a persistent session:
browser-use --session initial_test open https://example.com
browser-use-local Data Schema & Taxonomy
The skill manages various data outputs within the workspace to facilitate interaction with other Openclaw Skills. The organization is as follows:
| Asset Type | File Path / Pattern | Purpose |
|---|---|---|
| Session Screenshots | /home/node/.openclaw/workspace/*.png |
Visual debugging and element verification. |
| Raw HTML Exports | /tmp/page_html.json |
Temporary storage for deep link discovery and parsing. |
| QR Code Candidates | /home/node/.openclaw/workspace/qr_crops/ |
Extracted image regions for login automation. |
| Agent Logic | skills/browser-use-local/scripts/ |
Python implementation for Kimi and image processing. |
browser-use-local Advanced Features
- Support for persistent browser sessions to handle complex stateful interactions.
- Specialized compatibility flags for Moonshot/Kimi LLMs to bypass JSON schema validation errors.
- Automated image processing pipeline for identifying and cropping QR codes from login pages.
- Direct JavaScript evaluation capabilities for querying the DOM when standard tools return empty states.
- CLI-first architecture designed for seamless integration into larger automated scripts and pipelines.
SKILL.md
Loading
Related Openclaw Skills
China Demand Mining
An AI-powered market research agent that mines user demand and pain points from major Chinese social and e-commerce platforms to drive product innovation.
Smart Search
An automated search utility that intelligently toggles between Bing and DuckDuckGo based on language detection to provide the most relevant results.
Bing Search
A high-precision Microsoft Bing search integration providing filtered, ranked, and summarized web results without requiring an API key.
Openclw Translation Skill
A high-efficiency command-driven translation utility for Openclaw Skills that provides intelligent bidirectional Chinese-English translation and multi-language support.
Gemini Web Search
Perform deep web research and fact-finding with sourced summaries using the Gemini CLI.
China Hotel Comparison
A specialized AI agent skill designed for real-time price comparison and deep package value analysis across major Chinese travel platforms.
Featured*
