CONTENT / MARKDOWN
Markdown-for-Agents Checker
Request a page as HTML and Markdown. Check real content negotiation, then look for important headings, text and links missing from the converted response.
A PUBLIC PAGE / A USEFUL ANSWER
Check your page.
Your findings will appear here.
Run a check to see measured values, matched rules and specific next steps.
THE PROCESS
A few steps to a useful result.
- 01
Enter a public HTTPS page
Use the full URL of a page you want an agent to read. The checker fetches its public response without a login or JavaScript execution.
- 02
Compare the two responses
We request ordinary HTML and then Accept: text/markdown. A successful Markdown response must advertise the right content type; a 200 HTML response is not enough.
- 03
Review important content
Check headings, text and links against the original. Follow up on missing sections, then rerun after changing your conversion settings.
GOOD QUESTIONS
Before you use the results.
Is a .md file or llms.txt enough?
No. This check tests content negotiation on the actual page URL. An unrelated Markdown file or llms.txt does not prove the page responds to Accept: text/markdown.
Are token savings exact?
No. Cloudflare describes its token headers as estimates. Savings are shown only when valid original and Markdown token estimates are both present.
OFFICIAL GUIDANCE / REVIEWED 2026-10-02
Know where the rules come from.
Report results reflect the response available when you run the check. Operator requirements and private account settings can change.