A high-performance document parsing pipeline that converts complex PDFs and images into structured Markdown using GPU-accelerated OCR via Modal.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install ocr2markdown
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 ocr2markdown 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 OCR2Markdown?
OCR2Markdown is a specialized document parsing tool designed to bridge the gap between static documents and structured data. By leveraging the MinerU pipeline on remote L4 GPUs, this skill allows users to transform PDFs and images into high-quality Markdown while preserving original layouts, intricate tables, and mathematical formulas. This makes it an essential component for any workflow involving Openclaw Skills that requires precise document ingestion.
Built for efficiency, the skill offloads the heavy computational requirements of OCR and document analysis to the cloud using Modal. This architecture ensures that users can process large files or batches of documents without needing local high-end hardware, while still receiving perfectly formatted outputs ready for use in LLMs, documentation sites, or personal knowledge bases.
OCR2Markdown Use Cases
- Digitizing academic papers with complex mathematical formulas into LaTeX-compatible Markdown.
- Converting scanned business reports and multi-page tables into editable formats.
- Extracting structured text and images from legacy PDF archives for inclusion in AI knowledge bases.
- Batch processing image-based documents into searchable Markdown files via Openclaw Skills.
How OCR2Markdown Works
- The user identifies a specific PDF/image or a directory, and the skill assigns a unique task slug.
- Remote storage volumes are initialized on Modal to handle data transfer and model storage.
- Source files are uploaded to the remote cloud volume to prepare for GPU processing.
- The MinerU pipeline is executed on an L4 GPU, parsing the documents and extracting layout information.
- The processed results, including Markdown files and extracted images, are downloaded to the local directory while maintaining the original structure.
OCR2Markdown Setup
To use this within the Openclaw Skills ecosystem, ensure your environment meets these requirements:
- Verify Python 3.9+ is installed:
python -V. - Install the Modal CLI:
pip install modal. - Authenticate your Modal account:
modal setup. - Add the skill to your local environment using the Openclaw Skills manager to begin processing documents.
OCR2Markdown Data Schema & Taxonomy
The skill organizes processed data within a dedicated slug directory to ensure isolation and clarity. The output structure is as follows:
| File/Folder | Description |
|---|---|
<slug>/upload/ |
Contains the original source PDF or image files. |
<slug>/output/<stem>_ocr/ |
The root directory for the parsed results of a specific file. |
<stem>.md |
The primary Markdown output containing text and references. |
images/ |
A subdirectory containing all figures and images extracted from the document. |
*.json |
Metadata files containing layout information and model-specific data. |
OCR2Markdown Advanced Features
- Intelligent checkpointing that automatically skips already-processed PDFs to save time and credits.
- Support for remote L4 GPU acceleration, providing significantly faster throughput than CPU-based OCR.
- Robust layout preservation that accurately identifies headers, tables, and footers.
- Direct integration with Openclaw Skills allowing for seamless multi-agent document analysis workflows.
SKILL.md
Loading
Related Openclaw Skills
AI Video Editor Change Clothes
Automate professional-grade clothing swaps in video footage using simple text commands and cloud GPU rendering.
Text to Video Making AI
A powerful AI-driven video creation tool that converts text prompts and documents into professional-grade shareable videos in minutes.
Image to Video Free
A high-speed AI agent skill that transforms static JPG and PNG images into professionally animated 1080p video clips via simple text commands.
b402 Private DeFi
A privacy-preserving DeFi suite for AI agents that enables gasless swapping, lending, and cross-chain bridging through Zero-Knowledge proofs.
AI Video Editor Hugging Face
An automated, cloud-based video editing assistant that leverages open-source Hugging Face models to process raw footage via simple text commands.
Grafana Authoring and Operations
A secure AI agent skill for inspecting Grafana health, diagnosing dashboards, and managing alerting rules through a controlled, multi-stage workflow.
Featured*
