SuperBased provides OpenClaw agents with visual perception and manual desktop control through a suite of 72 MCP tools for capture, OCR, and GUI automation.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install superbased
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 superbased 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 SuperBased?
SuperBased is a powerful skill bundle designed to give Openclaw Skills both sensory input and physical agency within a desktop environment. It acts as the eyes and hands for AI agents, allowing them to capture screens, perform AI-driven vision analysis, and execute complex OCR tasks. Beyond simple perception, it enables full GUI automation, allowing agents to click, type, scroll, and drag just like a human user.
By integrating the SuperBased MCP server, developers can bridge the gap between LLM reasoning and local desktop execution. This skill is essential for workflows that require visual verification, interacting with legacy desktop applications, or automating multi-step UI processes that are not accessible via standard APIs.
SuperBased Use Cases
- Seeing and describing content on the user's screen in real-time.
- Verifying UI changes and performing visual regression testing.
- Reading long-form content hidden behind scrollable sections.
- Automating complex multi-step desktop workflows like form filling or menu navigation.
- Extracting text via OCR from images, PDF documents, or application windows.
- Handling in-flow CAPTCHA challenges with humanized cursor trajectories.
- Transcribing voice dictation and speech-to-text input directly into the agent workflow.
How SuperBased Works
- The user installs the SuperBased CLI and skill bundle within the Openclaw Skills ecosystem.
- The SuperBased MCP server is registered, exposing 72 specialized tools to the AI agent.
- When a visual or UI-based trigger occurs, the agent selects the appropriate sub-skill (e.g., screenshot, redact, or gui-automation).
- The agent executes tools via the MCP bridge to capture state or modify the desktop environment.
- Safety rails verify permissions and required confirmations before any state-modifying action is performed.
SuperBased Setup
To get started with SuperBased for Openclaw Skills, follow these steps:
# 1. Install this skills bundle from ClawHub
openclaw skills install superbased
# 2. Register the SuperBased MCP server
openclaw mcp set superbased '{"command":"superbased","args":["mcp"]}'
# 3. Install the SuperBased CLI globally
npm install -g superbased
Optional: Download the desktop app from superbased.app for a visual gallery and configuration interface.
SuperBased Data Schema & Taxonomy
SuperBased organizes its operations into specialized tool categories and manages local state via a dedicated directory.
| Category | Description | Key Tools |
|---|---|---|
| Capture | Screen and window image acquisition | superbased_screenshot, _window_list |
| AI & OCR | Visual reasoning and text extraction | superbased_ai, _ocr, _compress_text |
| GUI Automation | Humanized desktop interactions | _click, _type, _form_fill, _sequence |
| Recording | Visual QA and session history | _sessions, _diff, _baseline |
| Privacy | Security and redaction layers | superbased_redact, _annotate |
All session data and audit logs are stored locally in ~/.superbased/ for privacy and user auditing.
SuperBased Advanced Features
- Humanization V2: Choose between 'light', 'human', or 'paranoid' movement profiles to bypass bot detection.
- Visual Regression: Create baselines and perform image diffing to detect visual regressions in software builds.
- CAPTCHA Guidance: Advanced tools for solving drag puzzles, rotation puzzles, and image grids.
- Text Compression: Convert large 5K-token text blocks into token-efficient images for AI vision processing.
- Proactive Monitoring: Set the agent to watch for specific errors or UI states during long-running deployments.
SKILL.md
Loading
Related Openclaw Skills
Industry Research Skill
A professional-grade automated research tool that generates structured 10-dimension industry reports including price trends, supply chain mapping, and technical stock analysis.
XPComment Image Metadata Manager
An automated utility for managing XPComment metadata in image files using exiftool integration.
JF Open Pro Video Record
A professional Python-based skill for managing cloud and local video recordings from JFTech devices including playback, lists, and downloads.
Mailbird MCP
A powerful local Model Context Protocol server that allows AI agents to manage Mailbird email accounts, folders, and messages directly.
JF Open Pro PTZ Control
A professional Python developer tool for managing JF security camera movements, presets, and privacy masking via official API endpoints.
JF Open Pro Capture Livestream
A professional Python-based API tool for managing JF surveillance devices, enabling batch image capture and multi-protocol livestream retrieval.
Featured*
