A local, offline OCR solution that leverages Ollama vision models to extract text from images and screenshots without cloud API dependencies.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install ollama-ocr
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 ollama-ocr 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 Ollama OCR?
Ollama OCR is a specialized tool designed for privacy-conscious developers who need reliable optical character recognition directly on their local machines. By integrating with the Openclaw Skills ecosystem, this skill allows AI agents to interpret visual data using powerful vision models like glm-ocr and llama3.2-vision. It eliminates the need for expensive third-party OCR services, ensuring that sensitive data never leaves your infrastructure.
This skill is particularly effective for workflows requiring high-accuracy text extraction from screenshots, scanned documents, or digital photos. Whether you are building a personal knowledge management system or an automated data pipeline, Ollama OCR provides the bridge between raw image files and actionable text data within the Openclaw Skills framework.
Ollama OCR Use Cases
- Extracting text from system screenshots or application windows for documentation.
- Processing sensitive financial or personal documents where cloud-based OCR is prohibited.
- Automating Chinese text recognition using the specialized glm-ocr model.
- Offline data entry and digitization of receipts or physical notes in remote environments.
How Ollama OCR Works
- The AI agent receives an image file path and an optional model specification.
- The skill connects to a local Ollama instance via a pre-configured HTTP endpoint.
- The specified vision model (e.g., moondream or llava) processes the image data.
- The model performs visual analysis and converts identified characters into a text string.
- The extracted content is returned to the Openclaw Skills interface for the user or subsequent agent tasks.
Ollama OCR Setup
To get started with this skill, ensure you have Ollama installed and the necessary models pulled:
# Pull a vision model
ollama pull glm-ocr:latest
# Run the OCR script via CLI
python3 ollama_ocr.py /path/to/image.jpg glm-ocr:latest
Note: If you are running OpenClaw in Docker, the default endpoint is configured to http://172.17.0.2:11434. Adjust the OLLAMA_HOST variable in the script if your host configuration differs.
Ollama OCR Data Schema & Taxonomy
The skill manages model selection and output formatting based on the following model capabilities:
| Model | Primary Function | Estimated Size |
|---|---|---|
| glm-ocr:latest | Specialized Chinese OCR | 2.2GB |
| llava:7b | General Visual Analysis | 4.7GB |
| moondream | Fast, Lightweight OCR | 1.5GB |
| llama3.2-vision | High-performance Vision | 7GB+ |
All output is returned as a UTF-8 encoded string representing the text found within the processed image.
Ollama OCR Advanced Features
- Multi-model support allowing users to choose between speed (moondream) and accuracy (llama3.2-vision).
- Offline-first architecture designed for secure environments and low-latency local processing.
- Seamless Python API integration for embedding OCR capabilities into custom Openclaw Skills scripts.
- Docker-ready configuration for easy deployment in containerized AI agent stacks.
SKILL.md
Loading
Related Openclaw Skills
Omnimedia Content Production and Operations Expert
A professional-grade skill for AI agents to master the entire lifecycle of multi-platform content creation, distribution, and data-driven user operations.
Enable Chrome Gemini
A specialized utility to activate and repair the native Gemini (Glic) interface in Google Chrome across all major operating systems, regardless of regional restrictions.
Disposable Email
A powerful utility for programmatically creating temporary Mail.tm inboxes and extracting OTP codes or message content for automated workflows.
Tavily Search
An advanced suite of AI-optimized web search and research tools designed specifically for Openclaw Skills agents.
Safe Browsing Protection
A proactive security layer that identifies and blocks high-risk commands and sensitive data leaks during AI-driven web browsing.
Sanxin Form Filler
An intelligent document automation tool that maps report content into Word-based application forms while maintaining professional formatting standards.
Featured*
