📄 Markdown Header Splitter Actor

Split Markdown content into structured chunks using header hierarchy

This actor intelligently splits Markdown documents into semantically meaningful chunks based on header hierarchy. Built with LangChain's MarkdownHeaderTextSplitter, it preserves metadata and structure for downstream applications like RAG systems, documentation processing, and content analysis.

---

🚀 Key Features

✅ Header-Based Chunking
Split content at specified header levels (e.g., #, ##, ###).

✅ Metadata Preservation
Each chunk includes hierarchical header metadata for context tracking.

✅ Flexible Configuration

• Choose which headers to split on
• Strip headers from content or retain them
• Works with any Markdown content

✅ RAG-Ready Output
Chunks are formatted for direct use in vector databases or LLM pipelines.

---

📝 Use Cases

1. RAG Systems
   Split long documents into context-aware chunks for:

   • Vector database indexing
   • Context window optimization
   • Knowledge retrieval

2. Documentation Processing
   Break technical docs into sections for:

   • Summarization
   • Translation
   • Search indexing

3. Content Analysis
   Analyze/report on document structure (e.g., API reference parsing).

---

🛠️ Getting Started

Installation

1. Create an Apify account at apify.com
2. Navigate to Actor → Search for markdown-splitter
3. Run with Sample Input:

   1{
   2  "markdown_text": "# Title
   ## Section 1
   Content...
   ## Section 2
   More content...",
   3  "headers_to_split_on": ["#", "##"],
   4  "strip_headers": true
   5}
   

Input Parameters

Field	Type	Description
markdown_text	string	Markdown content to split (required)
headers_to_split_on	array	Header levels to split on (default: ["#", "##", "###", "####", "#####", "######"])
strip_headers	boolean	Remove headers from chunk content (default: true)

Output Format

1{
2  "chunks": [
3    {
4      "content": "Section content here",
5      "metadata": {
6        "Header 1": "Title",
7        "Header 2": "Section 1"
8      }
9    },
10    ...
11  ]
12}

---

📚 Example Workflow: RAG System Integration

1. Input: Technical documentation in Markdown
2. Process:
   • Split into sections using ## headers
   • Strip headers to focus on content
3. Output:
   • Chunks with hierarchy metadata → Embed with OpenAI/LLama
   • Store in your favorite vector database like : ChromaDB/Pinecone/Weaviate for retrieval