A sophisticated document recognition skill using PaddleOCR-VL and DeepSeek-OCR to extract structured data and text from complex Chinese documents and images.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install china-doc-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 china-doc-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 China Doc OCR?
China Doc OCR is a specialized tool designed for high-accuracy optical character recognition and structured data extraction from Chinese-language documents. By leveraging advanced models like PaddleOCR-VL-1.5 and DeepSeek-OCR, it provides developers with a seamless way to convert physical media into digital, editable formats. This skill is particularly optimized for the Chinese region, offering direct connections without requiring network proxies.
Whether you are dealing with scanned legal contracts, complex financial tables, or handheld photos of identification cards, this skill intelligently selects the best vision-language model to maintain structural integrity. It goes beyond simple text extraction by preserving Markdown hierarchies, tables, and bold formatting, making it an essential component of modern Openclaw Skills workflows.
China Doc OCR Use Cases
- Digitizing scanned PDF contracts into editable Markdown files.
- Automating data entry by extracting structured fields from VAT invoices and receipts.
- Recognizing information from official IDs, business licenses, and certificates.
- Converting complex data tables within images into CSV or Markdown formats for analysis.
- Extracting text from screenshots or low-resolution images for quick reference.
How China Doc OCR Works
- The user provides a file path or URL for a PDF, image (JPG, PNG, WebP), or base64 string.
- The skill analyzes the input type and user intent to select the appropriate prompt template (e.g., General OCR, Invoice Extraction, or Chart Analysis).
- It initiates a model selection strategy, prioritizing PaddleOCR-VL-1.5 for speed and accuracy, then falling back to DeepSeek-OCR or Qwen2.5-VL if needed.
- For multi-page PDFs, the system splits the document using pypdf and processes each page individually.
- The extracted content is formatted into the requested output type, such as structured Markdown or a data-rich schema.
- The final result is saved to the workspace for long-term retention and further processing.
China Doc OCR Setup
To use this skill within Openclaw Skills, ensure you have an API key from SiliconFlow.
# 1. Register at cloud.siliconflow.cn
# 2. Create your API Key in the dashboard
# 3. Export the key to your environment
export SILICONFLOW_API_KEY='sk-xxxxxxxx'
# 4. (Optional) Install pypdf for multi-page PDF support
pip install pypdf
China Doc OCR Data Schema & Taxonomy
China Doc OCR organizes its output based on the specific extraction target:
| Document Type | Output Format | Key Metadata Fields |
|---|---|---|
| General Documents | Markdown | Headers, Lists, Bold text, Code blocks |
| Invoices | Structured Text | Invoice Number, Date, Buyer/Seller, Tax Rate, Total Amount |
| Tables | CSV/Markdown | Column headers, Cell values, Row alignment |
| Identity Cards | Structured Text | Name, ID Number, Address, Validity Period |
| Charts | Analysis | Data trends, Axis labels, Summary statistics |
China Doc OCR Advanced Features
- Intelligent model fallback hierarchy to ensure successful recognition even with complex layouts.
- Direct support for high-resolution processing (up to 3584x3584 pixels) with automatic scaling.
- Specialized prompt templates for commercial documents like business licenses and VAT invoices.
- Integration with local workspace storage for persistent document management.
- Low-latency performance optimized for direct connection within mainland China.
SKILL.md
Loading
Related Openclaw Skills
Etalon GDPR Scan
A powerful privacy engineering tool that performs deep GDPR compliance audits on websites and codebases using a database of over 111,000 tracker domains.
Finality
A dedicated CLI utility for analyzing blockchain finality mechanisms and evaluating protocol security parameters.
Split: Data Partitioning & Splitting Reference
A comprehensive technical reference for data splitting, partitioning, and chunking strategies across software development and data science.
QSNCTF Practice Dynamic
A utility to fetch real-time leaderboard rankings and recent challenge solve activities from the QSNCTF platform.
Stock Theme Events
An automated analysis skill that identifies A-share market themes by correlating top-performing stocks with real-world news events.
HubSpot CRM
A comprehensive automation tool for managing HubSpot CRM contacts, deals, companies, and activities directly from the command line.
METADATA
Featured*
