A sophisticated document processing skill that converts images and PDFs into structured Markdown with layout preservation via the PaddleOCR API.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install paddleocr-doc-parsing-v2
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 paddleocr-doc-parsing-v2 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 PaddleOCR Document Parsing?
The PaddleOCR Document Parsing skill is a professional-grade tool designed to bridge the gap between static documents and structured data. By leveraging this entry in the Openclaw Skills library, developers can extract high-fidelity text, tables, and mathematical formulas from varied formats including PDF, JPG, PNG, and TIFF. It provides a seamless interface to the PaddleOCR engine, ensuring that document structure is maintained throughout the conversion process.
Whether you are building a RAG pipeline or automating data entry, this skill offers the reliability of a battle-tested OCR engine with the flexibility of both synchronous and asynchronous processing. It is specifically optimized to handle everything from single-page screenshots to massive multi-page PDF reports with ease.
PaddleOCR Document Parsing Use Cases
- Converting legacy PDF archives into searchable Markdown for AI knowledge bases.
- Extracting data from complex financial tables and reports with high structural accuracy.
- Processing large batches of scanned documents using asynchronous polling for high throughput.
- Rapidly digitizing research papers and technical manuals while preserving formulas and layouts.
How PaddleOCR Document Parsing Works
- The user initiates a request by passing a file path or URL to the script.
- The system authenticates with the PaddleOCR API using provided environment credentials.
- Based on file size and user preference, the skill selects between Synchronous (real-time) or Asynchronous (job-based) processing.
- The PaddleOCR engine performs deep layout analysis to identify text blocks, tables, and formatting elements.
- The raw API response is transformed into a structured JSON object containing both metadata and a clean Markdown representation.
- Results are either printed to the console or saved to a specified output file for further integration.
PaddleOCR Document Parsing Setup
To utilize this skill within the Openclaw Skills framework, you must first obtain API credentials from PaddleOCR. Then, configure your environment variables as follows:
export PADDLEOCR_ACCESS_TOKEN="your_token_here"
export PADDLEOCR_API_URL="https://your-endpoint.aistudio-app.com/layout-parsing"
# Optional: Configuration for large files (Async Mode)
export PADDLEOCR_JOB_URL="https://your-job-endpoint.aistudio-app.com/api/v2/ocr/jobs"
export PADDLEOCR_MODEL="PaddleOCR-VL-1.5"
PaddleOCR Document Parsing Data Schema & Taxonomy
The skill returns a comprehensive JSON structure that captures every detail of the document layout:
| Key | Description |
|---|---|
logId |
A unique identifier for the API request used for tracking and debugging. |
markdown.text |
The core document content, fully formatted in Markdown. |
prunedResult |
Detailed information on layout elements, including bounding boxes and categories. |
outputImages |
A list of images extracted from the original document (e.g., charts, photos). |
dataInfo |
Metadata regarding the input file processing state and document statistics. |
PaddleOCR Document Parsing Advanced Features
- Intelligent Layout Analysis: Automatically detects and formats tables and mathematical formulas into Markdown.
- Dual Processing Modes: Switch between Sync mode for instant feedback and Async mode for heavy-duty batch processing.
- Extensive Language Support: Capable of recognizing text in over 110 different languages.
- CLI Versatility: Includes both shell (
.sh) and Python (.py) entry points for maximum compatibility across environments.
SKILL.md
Loading
Related Openclaw Skills
Reddit Spy
A stealth intelligence tool for analyzing subreddits, user patterns, and content strategies without detection or blocking.
PrepSPSC PYQ
A professional-grade API skill for searching 7,400+ Sikkim PSC previous year questions and generating difficulty-balanced mock tests for Indian government exams.
MiniMax MCP Search Skill
A powerful search and vision skill for AI agents using MiniMax MCP to fetch live web data and analyze images.
Nextbrowser
Spin up cloud-based browsers with persistent profiles and residential proxies for autonomous web automation.
Timer Skill
A powerful background utility for setting precise timers, alarms, and reminders through your AI coding agent.
Withings Health Integration
A comprehensive health data retrieval skill that connects AI agents to the Withings API for weight, activity, and sleep analysis.
Featured*
