An advanced document understanding skill powered by IBM's docling for parsing complex layouts, tables, and figures.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install doc-parser
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 doc-parser 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 Document Parser (Docling)?
The Document Parser is a high-performance tool designed for deep document understanding and structural analysis. By utilizing IBM's docling library, this skill allows users to convert complex PDFs, Word documents, and images into structured, machine-readable formats like Markdown, JSON, and plain text. As a core part of the Openclaw Skills library, it provides the necessary infrastructure to handle sophisticated layout challenges that traditional parsers often fail to process.
Whether you are dealing with academic papers featuring multi-column layouts or financial reports packed with nested tables, this skill ensures that the hierarchy and context of the information remain intact. It is an essential utility for developers looking to build robust RAG (Retrieval-Augmented Generation) pipelines or automated data extraction workflows using Openclaw Skills.
Document Parser (Docling) Use Cases
- Extracting and converting financial tables from annual reports into Pandas DataFrames for analysis.
- Transforming academic manuscripts with multi-column layouts into clean Markdown for LLM training.
- Automating the extraction of technical diagrams and their associated captions from engineering manuals.
- Batch processing legacy PDF archives into structured JSON to populate searchable databases.
- Parsing business contracts to identify key clauses, dates, and monetary amounts systematically.
How Document Parser (Docling) Works
- The skill initializes a DocumentConverter instance, which can be configured with specific pipeline options like OCR or table structure recognition.
- The input file (PDF, DOCX, PPTX, HTML, or Image) is ingested and processed by the docling engine.
- A sophisticated layout analysis is performed to identify structural elements including headings, paragraphs, tables, and figures while respecting the natural reading order.
- The document is converted into an internal document model that maintains the hierarchy of the original content.
- Users can then query the model to export specific segments, such as dataframes for tables or PNG files for figures, or export the entire document into a chosen format like Markdown via Openclaw Skills.
Document Parser (Docling) Setup
To begin using this skill, you must install the docling library and its required dependencies in your environment:
# Install the base docling package
pip install docling
# Install with all features (recommended for maximum compatibility)
pip install docling[all]
# Install specifically for OCR and scanned document support
pip install docling[ocr]
Document Parser (Docling) Data Schema & Taxonomy
The Document Parser organizes extracted data into a hierarchical structure that maps directly to the original document's layout. Below is how the schema is generally organized:
| Component | Format | Metadata Collected |
|---|---|---|
| Core Document | docling object | Name, origin source, and document hash |
| Tables | Pandas DataFrame | Page number, bounding box, and cell coordinates |
| Figures/Pictures | Image (PNG) | Associated captions and page location |
| Text Content | Markdown/JSON | Heading levels, paragraph hierarchy, and reading order |
| Metadata | Dictionary | Title, author, and date (where available) |
Document Parser (Docling) Advanced Features
- Multi-column layout detection to ensure text is extracted in the correct logical reading order.
- Advanced table structure recognition with cell-matching capabilities for complex financial data.
- Built-in OCR support for handling scanned documents and image-based PDFs within Openclaw Skills.
- Parallel batch processing using ThreadPoolExecutor for high-throughput document conversion.
- Flexible pipeline configuration to toggle specific extraction features like figure saving or table analysis.
- Native export support for multiple formats including Markdown, JSON, and plain text.
SKILL.md
Loading
Related Openclaw Skills
Developer Slides (dev-slides)
An AI skill for generating interactive, developer-focused presentations using the Slidev framework and Markdown.
Data Extractor
Extract and structure data from PDFs, Word docs, and emails using unified document processing.
CV Builder
A professional resume generation skill that converts structured YAML data into polished PDF CVs using multiple industry-standard themes.
Chat with PDF
An advanced AI skill for intelligent conversations, data extraction, and automated summarization of PDF documents.
DOCX Manipulation
A powerful skill for programmatically creating, editing, and formatting Microsoft Word documents using the python-docx library.
Excel Automation with xlwings
Advanced Excel automation using Python and xlwings to interact with live spreadsheet instances and execute complex workflows.
Featured*
