When building RAG pipelines in ODC against structured documents, the four native chunking methods treat all input as plain text. This works well for simple text stored in entity attributes but creates problems for documents that have structure: fenced code blocks, Markdown tables, and nested headings.

A Markdown-aware chunking option would let developers preserve that structure during ingestion without needing a custom External Logic workaround. Three behaviours would cover most cases: heading path tracking so chunks carry their section context, code block atomicity so fenced blocks are never split across chunks, and table preservation so tables stay whole.

The upstream Forge ecosystem already produces Markdown output via components like OmniDoc2MD. A native Markdown chunking method would complete that pipeline naturally.

This is not a replacement for the existing methods. It is an additional option for teams whose documents have structure worth preserving.