An intelligent image analysis tool for Openclaw Skills that performs OCR, object detection, and detailed visual descriptions in any language.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install vision-bot
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 vision-bot 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 Vision Bot?
Vision Bot is a versatile image processing skill designed to seamlessly integrate visual intelligence into your automated workflows. As a core component of the Openclaw Skills ecosystem, it allows developers and agents to analyze images, extract text via OCR, and detect objects simply by providing a URL within a task string.
This skill intelligently adapts to your specific needs, whether you are asking for a general description, a count of specific items, or comprehensive text extraction from a screenshot. By leveraging high-performance vision models, Vision Bot ensures that your AI agents can see and interpret the physical or digital world with high accuracy and speed.
Vision Bot Use Cases
- Generating accessibility descriptions for web images and digital assets.
- Extracting text from photos, signs, or software UI screenshots using high-precision OCR.
- Auditing webpage screenshots for visual consistency, content verification, and design reviews.
- Performing object detection and counting for inventory management or data analysis.
- Analyzing visual data across multiple languages including Spanish, Chinese, French, and more.
How Vision Bot Works
- The user provides a natural language task string containing an image URL or provides the image_url field separately.
- Vision Bot parses the task context to determine the required analysis mode, such as OCR, counting, or descriptive analysis.
- The skill establishes a secure network connection to the orchestration endpoint using the AIPROX_SPEND_TOKEN for authentication.
- Claude vision capabilities process the image transiently to ensure data privacy and high reasoning quality.
- The skill returns a structured JSON response containing the visual description, a list of detected objects, and any discovered text in the language of the original task.
Vision Bot Setup
To deploy Vision Bot within your Openclaw Skills environment, you must configure the required spending token for API access.
export AIPROX_SPEND_TOKEN="your_token_here"
You can then make requests directly via the orchestration API:
curl -X POST https://aiprox.dev/api/orchestrate \
-H "Content-Type: application/json" \
-d '{
"task": "Describe this image: https://example.com/photo.jpg",
"rail": "bitcoin-lightning",
"spend_token": "$AIPROX_SPEND_TOKEN"
}'
Vision Bot Data Schema & Taxonomy
Vision Bot returns data in a clear, structured format to allow for easy downstream processing by other Openclaw Skills.
| Property | Type | Description |
|---|---|---|
description |
string | A detailed narrative description of the image content. |
objects |
array | A list of specific items or entities identified in the image. |
text_found |
string | Extracted text found within the image via OCR. |
Vision Bot Advanced Features
- Automatic language detection to ensure response language matches the input task.
- Intelligent mode switching that detects if a user is asking for text extraction or counting without manual configuration.
- Support for both direct URL parsing from task strings and dedicated image_url fields.
- High-performance transient processing that ensures no images are stored permanently, maintaining high security standards for Openclaw Skills users.
SKILL.md
Loading
Related Openclaw Skills
Publish Checklist
A rigorous pre-release review protocol designed to prevent data leaks and ensure security when publishing code, PRs, or public content.
Data Spider
A high-performance web scraping agent that extracts structured data from any webpage using AI-driven schema mapping.
JournalBot: Daily AI Journaling
A private, local AI-driven journaling assistant that provides smart prompts and analyzes emotional patterns without cloud dependency.
Copywriting
An expert conversion copywriting skill designed to produce clear, compelling, and action-oriented marketing copy for web pages and product offers.
DeepL
A powerful integration for Openclaw Skills that enables neural machine translation and advanced text rephrasing via the DeepL API.
Brandomica Lab
A comprehensive brand verification engine to check name safety across domains, social handles, trademarks, and package registries.
Featured*
