General Ocr Struct

What This Skill Does

The General OCR Struct skill provides a robust, offline-first optical character recognition (OCR) engine for OpenClaw. Powered by RapidOCR, this skill specializes in transforming raw pixels from screenshots, scans, or images into usable text data. Unlike generic tools, it separates the recognition process from the interpretive phase, ensuring that raw data is captured reliably before any downstream processing (like summarization or data entry) occurs. It is particularly adept at handling mixed-language environments, such as combined Chinese and English text, as well as complex layouts like tabular transaction data and receipts.

Installation

To integrate this skill into your OpenClaw environment, execute the following command in your terminal:

clawhub install openclaw/skills/skills/9penny/general-ocr-struct

Ensure that you have a compatible Python environment installed. The first time you execute the command, the skill may trigger a one-time dependency installation process to fetch necessary libraries. Once initialized, the skill runs entirely offline, protecting your data privacy and ensuring zero latency from external API calls.

Use Cases

This skill is designed for scenarios where accuracy and data integrity are paramount:

• Financial Reconciliation: Extract transaction details from bank statements or e-wallet screenshots, organizing them into machine-readable formats.
• Content Digitization: Convert images of scanned documents, meeting minutes, or handwritten notes into editable digital text.
• Workflow Automation: Capture data from recurring reports or chat screenshots for automatic insertion into Excel or database schemas.
• Verification: Act as a middle-ware layer where raw OCR text is reviewed by the user before an AI agent generates a summary or makes a business decision based on the content.

Example Prompts

1. "Run raw OCR on the screenshot in my downloads folder and show me the text output so I can check for errors."
2. "Extract the transaction rows from this bank screenshot and give me the output in JSON format so I can import it into my spreadsheet."
3. "Please process this image of a receipt, group the items by price and description, and mark any blurry parts as 待确认."

Tips & Limitations

For the best results, always ensure that your input images are of high resolution. If the OCR engine struggles to identify characters, the text output may be fragmented or incorrect. Avoid inferring data when the text is illegible; the skill is configured to use '待确认' (To be confirmed) to maintain data accuracy. By following the workflow of 'Recognize -> Review -> Structure', you prevent the propagation of AI hallucinations in your downstream data pipelines.