I was burning context feeding HTML to LLMs, so I built a Markdown converter for agents

If you build AI agents, scrapers, or RAG pipelines, you have run into this: feeding raw HTML to a...

giovedì 25 giugno 2026 New tab

622 words~3 min read

If you build AI agents, scrapers, or RAG pipelines, you have run into this: feeding raw HTML to a model is mostly waste. The <div>, the class="...", the nav, the cookie banner. None of it is the content you wanted, and all of it costs tokens.

I knew this was inefficient. I did not appreciate how inefficient until I measured it. Here is a single page of GitHub's documentation, run through an audit:

HTML Markdown Savings

───────────────────────────────────────────────────

Tokens 138,550 9,364 -93.2%

I was burning context feeding HTML to LLMs, so I built a Markdown converter for agents

I was burning context feeding HTML to LLMs, so I built a Markdown converter for agents

Related reading

Reduce LLM Token Waste in RAG with Markdown

How to Convert Webpages into Clean Markdown for LLMs (in 5ms)

Markdown Is the Operating System. Everything Else Is a Render.

We benchmarked an 84% token reduction. Then we open sourced the protocol.

Your AI Agent Is Paying for HTML It Never Reads — I Measured the 7x Token Tax

How Our Document Ingestion Pipeline Turns Files into LLM-Ready Markdown

Related reading

Reduce LLM Token Waste in RAG with Markdown

How to Convert Webpages into Clean Markdown for LLMs (in 5ms)

Markdown Is the Operating System. Everything Else Is a Render.

We benchmarked an 84% token reduction. Then we open sourced the protocol.

Your AI Agent Is Paying for HTML It Never Reads — I Measured the 7x Token Tax

How Our Document Ingestion Pipeline Turns Files into LLM-Ready Markdown