skills.rest
orbruno
orbrunoCOMMUNITY
Explore all skills by this author →
GitHub

Docling Chunking

Extract structure-aware chunks with page and heading metadata for RAG workflows.

1|1|Updated Feb 4, 2026
Search Tags:#pdf#embeddings#rag#metadata#structure#chunking#docling
One-click install
npx skills add https://github.com/orbruno/docling-ccplugin --skill docling-chunking
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: Docling Chunking
Source: https://github.com/orbruno/docling-ccplugin/tree/main/skills/docling-chunking
Command: npx skills add https://github.com/orbruno/docling-ccplugin --skill docling-chunking

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve?

Docling Chunking addresses the difficulty of extracting meaningful, navigable chunks from long documents while preserving document structure, provenance, and metadata for robust retrieval and analysis.

Core Features & Use Cases

  • HybridChunker provides balanced structure-aware chunking optimized for embeddings and RAG workflows.
  • HierarchicalChunker preserves exact document structure for precise citations and hierarchical knowledge representations.
  • Rich per-chunk metadata (pages, section headings, provenance) supports high-quality search, auditing, and traceability.
  • Export options enable JSONL and Markdown outputs for ingestion by vector stores, databases, and knowledge bases.
  • Use cases include building document QA pipelines, knowledge bases, and compliant citation trails across multi-document corpora.

Quick Start

Run the example workflow to convert a document and generate hierarchical chunks ready for RAG pipelines.

Frequently Asked Questions about Docling Chunking

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I preserve PDF structure and page numbers for RAG chunking?▼

Structure-aware chunking extracts and preserves document hierarchy, section headings, and page numbers from PDFs to ensure reliable RAG retrieval. It outputs provenance-rich metadata so each chunk is traceable to its exact origin.

What is the best way to chunk long documents for vector store ingestion?▼

HybridChunker provides balanced structure-aware chunking optimized for embeddings, while HierarchicalChunker preserves exact document structure. Both export JSONL or Markdown outputs ready for vector stores, databases, and knowledge bases.

Can I extract hierarchical chunks with section headings for document QA pipelines?▼

Yes, HierarchicalChunker preserves exact document structure for precise citations. It enables hierarchical knowledge representations across multi-document corpora, making it suitable for building compliant document QA pipelines with auditable citation trails.

Does this chunking method work with HTML and other formats besides PDF?▼

Structure-aware chunking applies to PDFs, HTML, and other supported formats. It extracts navigable chunks while preserving metadata, ensuring robust retrieval and provenance-rich JSONL or Markdown exports for downstream AI tasks.

Why do I need provenance metadata like page numbers in my RAG chunks?▼

Provenance metadata like page numbers and section headings supports high-quality search, auditing, and traceability. It ensures rigorous citation trails across multi-document corpora, allowing reliable verification of retrieved information.

What is the difference between HybridChunker and HierarchicalChunker for embeddings?▼

HybridChunker provides balanced structure-aware chunking optimized for embeddings and RAG workflows, whereas HierarchicalChunker preserves exact document structure for precise citations and hierarchical knowledge representations.

Related Skills

Explore More Data & Analytics Skills→
jeffvincentjeffvincent
5

UXR Report Analyzer

Extract UXR report PDFs into standardized Markdown analyses with metadata.

Community
Advanced
patrykkopycinskipatrykkopycinski
6

rag-pipeline

Set up a RAG pipeline with Elasticsearch for vector retrieval.

Community
Intermediate
adeboldadebold

AgentDB Vector Search

Perform vector-based semantic search against an AgentDB index.

Community
Advanced
evanfang0054evanfang0054

langchain-text-splitters

Split large documents into smaller chunks using LangChain.js text splitters.

Community
Intermediate