A professional OCR skill for high-speed, offline recognition of invoices, train tickets, and general receipts.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install rapid-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 rapid-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 RapidOCR Skill?
RapidOCR Skill is a specialized tool within the Openclaw Skills ecosystem designed for professional-grade document and receipt recognition. It focuses on high-speed extraction from value-added tax invoices, railway tickets, and other financial documents. Built with a lightweight footprint of only 4KB, it prioritizes speed and developer transparency.
The skill is uniquely engineered for privacy and efficiency. After a one-time automatic model download of approximately 13MB during the first run, the tool operates entirely offline. This ensures that sensitive financial data processed via Openclaw Skills never leaves the local environment, providing a secure alternative to cloud-based OCR services.
RapidOCR Skill Use Cases
- Automating expense reports by extracting data from digital receipts.
- Digitizing physical train tickets and travel documents for record-keeping.
- Processing high volumes of value-added tax invoices for accounting software integration.
- Building privacy-centric local tools that require text extraction without cloud API costs.
How RapidOCR Skill Works
- The user provides an image path (JPG/PNG) via the CLI or Python API.
- On the initial execution, the skill triggers the rapidocr-onnxruntime dependency to download the necessary PaddleOCR models (~13MB).
- The image is processed locally using the ONNX runtime engine for high-performance inference (~500ms per image).
- The skill parses the raw OCR results into structured data, specifically identifying fields for invoices or general text.
- The final output is returned as a JSON-compatible dictionary containing full text or structured key-value pairs.
RapidOCR Skill Setup
To use this in your environment, ensure you have Python installed and install the required dependencies:
pip install rapidocr-onnxruntime
You can then run the skill directly from the terminal:
# General OCR
python rapidocr_minimal.py ocr image.jpg
# Invoice-specific extraction
python rapidocr_minimal.py invoice invoice.jpg
# Train ticket extraction
python rapidocr_minimal.py train ticket.jpg
RapidOCR Skill Data Schema & Taxonomy
The RapidOCR Skill provides structured output to ensure compatibility across Openclaw Skills workflows.
| Field | Type | Description |
|---|---|---|
full_text |
String | The raw concatenated text recognized from the image. |
structured_data |
Object | Key-value pairs (e.g., Date, Total, Invoice Number) for specific receipt types. |
confidence |
Float | The accuracy score of the OCR inference. |
metadata |
Object | Image dimensions and processing time. |
RapidOCR Skill Advanced Features
- Offline Inference: Operates without any runtime network requests to guarantee data sovereignty.
- Low Latency: Optimized for speed with an average processing time of 500ms per document.
- Developer-Friendly API: Seamlessly integrates into larger Python projects as a modular class.
- Secure Auditing: No use of risky functions like exec() or eval(), making it safe for enterprise Openclaw Skills deployments.
SKILL.md
Loading
Related Openclaw Skills
Lab-to-Market Co-Creation Method
A comprehensive framework for university hard-tech transfer that integrates Tsinghua co-creation models with MIT's 24-step disciplined entrepreneurship.
Reflectt
A comprehensive coordination and operator layer for AI agents to manage tasks, communicate in team chats, and maintain operational visibility.
FTShare K-Line Data
A comprehensive A-share market data tool for retrieving historical K-line patterns and minute-by-minute price movements.
Stock Research Team
A sophisticated multi-agent system that orchestrates technical, fundamental, and macro analysts to provide comprehensive stock market research and trading strategies.
Pilates (Session Planner, Form Coach, Progress Tracker)
A comprehensive AI-driven assistant for planning Pilates sessions, refining alignment, and tracking long-term core progression.
Booking Hotel Search
A powerful hotel discovery skill that integrates real-time Booking.com data into your AI agent for instant availability and price comparisons.
Featured*
