Markdown Browser (markdownbrowser.com) is a web tool developed by Jan Wilmake and launched on March 2, 2026. It shows any URL's markdown representation alongside a llms.txt navigation sidebar — framed as a browser for the "second web" of AI-readable content.[1]
Jan's launch framing: "The web is bifurcating. For 30 years we had one web — built for human eyes. Now a second web is emerging: markdown, structured data, and llms.txt files that AI agents consume. There's no browser for that second web. Until now."[1]
The core observation motivating Markdown Browser: "Markdown is a better reading experience. No popups. No layout shift. No cookie dialogs. No autoplaying videos. No 'subscribe' modals. Just text, links, and code blocks. It's reader mode on steroids — except the content is designed to be read this way."[1]
Sites with /llms.txt files become navigable through a sidebar, similar to a dev portal but for the machine-readable web. Devtools show token cost, response time, and spec compliance.
markdownbrowser.com/?url=https://...[2]| Date | Milestone |
|---|---|
| Mar 2, 2026 | Initial launch; announcement thread on X |
| Mar 6, 2026 | Added pluggable extract and search engines; Defuddle becomes default |
| Mar 6, 2026 | Added deep link sharing via URL query param |
For URLs that only serve HTML (no native markdown), Markdown Browser requires a Parallel API key to use Parallel's extraction API to render them. Sites that natively serve markdown (via content-negotiation) work without a key.
Jan positioned this as a showcase for Parallel's Extract API — the right extract engine matters, and Parallel's is one of the best options.[2]
The launch thread received positive engagement. Jan noted that some sites "hold up beautifully" in markdown while others don't, surfacing which sites are genuinely agent-friendly.