document-chunking

Chunk long documents into structured blocks with heading context.

13|4|Updated Apr 27, 2026
One-click install
npx skills add https://github.com/memect/kc --skill document-chunking
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: document-chunking
Source: https://github.com/memect/kc/tree/main/template/skills/zh/document-chunking
Command: npx skills add https://github.com/memect/kc --skill document-chunking

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Breaks long documents into structured blocks for downstream processing, enabling efficient review, extraction workflows, and batch processing of regulatory texts.

Core Features & Use Cases

  • Cheap chunking methods: by page, fixed-size, or by title to rapidly explore data.
  • Hierarchical onion-peeling chunking for long documents, with a fallback wedge-in approach when structure is absent.
  • Context preservation: keeps heading chains with each block to maintain provenance and navigation.
  • Production-ready outputs suitable for downstream agents and workflows.

Quick Start

Provide a sample document to chunk and obtain structured blocks ready for downstream analysis.

Frequently Asked Questions about document-chunking

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

FAQPage Schema
How do I chunk long regulatory documents while preserving heading structure?

Chunk long regulatory documents using a hierarchical onion-peeling strategy that preserves parent heading chains within each block, maintaining provenance and navigation context for downstream extraction workflows.

What is the best way to break unstructured long documents into blocks for extraction?

Breaking unstructured long documents into blocks uses a wedge-in fallback approach when document structure is absent, ensuring continuous block output suitable for downstream batch processing and extraction tasks.

Can I chunk documents by page, fixed size, or title for rapid data exploration?

Yes, you can chunk documents by page, fixed size, or title to rapidly explore data and generate production-ready blocks for immediate downstream analysis without complex setup.

Does structure-preserving chunking keep heading context for downstream agents?

Structure-preserving chunking keeps the parent heading chain attached to each block, ensuring downstream agents receive full contextual provenance for accurate extraction from long regulatory texts.

What are the limitations of cheap chunking methods for long document processing?

Cheap chunking methods like page, fixed-size, or title splitting offer rapid exploration but lack deep structural awareness, making them less suitable than onion-peeling for complex hierarchical regulatory documents.