Prepare HTML Content for LLM Prompts

LLMs handle Markdown better than HTML — fewer tokens, cleaner structure. Convert HTML before pasting context into prompts.

For RAG (retrieval augmented generation) systems: store source documents as Markdown for cleaner chunking and embedding.

Token savings: a 5000-token HTML document often becomes a 3000-token Markdown version — leaves more room for the actual question.

Strip code blocks separately if you need the LLM to focus on text. The Markdown output keeps `code` fenced for easy identification.