A high-performance, offline OCR tool for structured data extraction from invoices, travel tickets, and medical receipts.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install fast-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 fast-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 FastOCR?
FastOCR is a professional-grade optical character recognition tool designed specifically for the structured extraction of data from various financial and travel documents. Built on the ONNX inference engine, it offers a lightweight and fast solution that operates entirely offline, ensuring maximum data privacy for sensitive financial information within the library of Openclaw Skills.
This skill excels at identifying complex layouts, from standard VAT invoices to high-speed rail tickets, converting visual data into actionable JSON objects. By eliminating the need for heavy deep learning frameworks, it provides developers and AI agents with a robust tool for automating accounting, travel management, and general text recognition tasks without sending data to external servers.
FastOCR Use Cases
- Automated financial reimbursement and expense report entry for accounting teams.
- Batch processing of corporate receipts to export data into Excel or CSV formats.
- Travel itinerary tracking and expense calculation via train and flight ticket analysis.
- Medical insurance claim processing through structured medical invoice extraction.
- General productivity workflows like extracting text from screenshots or digitized paper documents.
How FastOCR Works
- The user provides an image path or a directory containing receipts to the AI agent.
- FastOCR identifies the document type, such as a VAT invoice, train ticket, or taxi receipt.
- The local ONNX inference engine processes the image to perform high-precision character recognition.
- The system automatically maps recognized text into a structured schema with over 30 specific metadata fields.
- The skill returns a structured JSON object containing the extracted data and confidence scores.
FastOCR Setup
To use this tool within Openclaw Skills, you must first install the required ONNX runtime:
pip install rapidocr-onnxruntime
For users requiring faster performance on large batches, GPU acceleration can be enabled:
pip install onnxruntime-gpu
FastOCR Data Schema & Taxonomy
FastOCR organizes extracted data into specialized JSON structures based on the document type. Below is the primary schema taxonomy:
| Data Group | Included Fields |
|---|---|
| Basic Info | Invoice code, number, date, check code, and document type |
| Parties | Buyer/Seller names, Tax IDs, addresses, phone numbers, and bank details |
| Financials | Pre-tax amount, tax rate, tax amount, and total amount with tax |
| Travel Specifics | Train/flight numbers, seat class, carriage, and departure/arrival stations |
| General OCR | Full text string, individual lines, and recognition confidence levels |
FastOCR Advanced Features
- Support for 100% offline operation to maintain strict data privacy and security standards.
- High-speed inference engine delivering results in approximately 300-500ms per document.
- Batch processing mode for handling entire directories of images in a single command.
- Configurable confidence thresholds to flag low-quality scans for human review.
- Cross-platform support for Windows, macOS, and Linux environments.
SKILL.md
Loading
Related Openclaw Skills
JoinQuant Quantitative Trading
A comprehensive quantitative trading skill providing deep access to China's A-share, futures, and fund markets for automated research and strategy execution.
Word Chinese Format Standardizer
An automated utility to standardize Word documents according to professional Chinese typesetting norms, covering fonts, headings, and visual elements.
Soul MD Newsletter Connector
A streamlined RSS automation skill designed to track, fetch, and parse the latest editions of the Soul MD newsletter for AI agents.
EE Datasheet Master
A specialized technical analysis tool for extracting high-fidelity electrical specifications, pinouts, and register maps directly from electronic component datasheets.
Membox Cloud Sync
A private, zero-knowledge encrypted sync and recovery skill for maintaining AI agent memory across multiple devices.
Substreams Search
A comprehensive toolset for discovering, visualizing, and configuring Substreams packages from the substreams.dev registry.
Featured*
