Markdown content negotiation

What it checks

The extension requests the page being scanned (not the homepage) with the Accept header a real agent sends:

Accept: text/markdown, text/html;q=0.9, */*;q=0.8

It passes when the response is successful, has a text/markdown (or text/x-markdown) content type, and the body does not start like an HTML document. An error page labelled as Markdown does not count.

If that fails, the extension tries two fallbacks so it can tell you what is close to working:

• A Markdown-looking body served as text/plain, which differs from what the same URL returns for Accept: text/html.
• Markdown served only when the request sends a bare Accept: text/markdown, which no real client sends.

This is the heaviest check in Content accessibility. The .md URL and rel="alternate" routes to Markdown are separate checks: Markdown URL fallback and Markdown alternate link.

Results

Status	When
Pass	The realistic Accept header gets a successful text/markdown response
Warn	A Markdown body is returned, but labelled text/plain
Warn	Markdown is returned only for a bare Accept: text/markdown, not the realistic header
Fail	No Markdown version is offered through content negotiation

How to fix

When a request’s Accept header prefers text/markdown, respond with a Markdown version of the page and Content-Type: text/markdown. Parse the header properly, including quality values, rather than matching the exact string text/markdown.

Also send Vary: Accept, so caches keep the HTML and Markdown responses apart (see Markdown negotiation caching).

To test it:

curl -sI -H 'Accept: text/markdown, text/html;q=0.9, */*;q=0.8' https://example.com/page

The response should include Content-Type: text/markdown.

Support by agent and browser

Announced means the vendor’s own docs or announcement say so. Published is the date on the vendor’s own page, where it has one. Checked is the last time our weekly research found that page still saying so.

Agent	Status	Source
Coding agents
Claude Code Anthropic	Supported (announced)	Cloudflare blog: Introducing Markdown for Agents Published 12 Feb 2026 · Checked 9 Oct 2026Cloudflare (not Anthropic) states: 'some of the most popular coding agents today – like Claude Code and OpenCode – send these accept headers'. A request to httpbin.org/headers on 2026-10-09 confirmed it directly: Claude Code 2.1.291 WebFetch sent 'Accept: text/markdown, text/html, */*' with User-Agent 'Claude-User (claude-code/2.1.291; ...)'. Anthropic's own docs do not document the header.
GitHub Copilot GitHub	Supported (announced)	microsoft/vscode commit f6b248c: Enhance WebPageLoader for markdown content extraction (#320146) Published 11 Jun 2026 · Checked 9 Oct 2026This commit added Accept 'text/markdown, text/html;q=0.9, application/xhtml+xml;q=0.9, application/xml;q=0.8, */*;q=0.7' for main-frame requests, with the comment 'prefer markdown responses from sites that support agent-friendly content negotiation'. When the response is text/markdown it is used directly. It applies to the VS Code fetch tool, in VS Code releases after the merge. Checkly's Feb 2026 test, before this change, found Copilot sending an HTML Accept header.

No evidence yet25

• ChatGPT
• Claude
• Gemini
• Perplexity
• Copilot
• Meta AI
• ChatGPT Atlas
• Comet
• Dia
• Gemini in Chrome
• Edge Copilot Mode
• Cursor
• Codex
• Windsurf
• CCBot
• Applebot
• Amazonbot
• Bytespider
• Cohere
• Mistral
• DuckAssist
• Chrome
• Edge
• Firefox
• Safari

Names and marks belong to their owners and are used only to identify products. None of these vendors is affiliated with Sitebulb. Spotted something out of date? Send a correction.

Adoption on the top 100 sites

2%

2 of 99 measured sites implement it

Scanned 23 Oct 2026 with the Sitebulb Labs extension, against the Tranco list (tranco-56Z8N). 1 sites couldn’t be measured (bot challenge or timeout) and are left out of the share.

Sites implementing it2

1. #2 cloudflare.com
2. #55 wordpress.org

Overview

6/9

Gaining traction

Real support exists and adoption is growing.

How this rating is worked out6 of 9 points

Factor	Points	Why
Vendor support	2 / 3	2 vendors supporting, 0 partly or planned
Top-100 adoption	1 / 3	2% of measured sites
Spec maturity	3 / 3	Published standard

8 or more points is Established, 5 to 7 is Gaining traction, 2 to 4 is Speculative and under 2 is Stalled. A feature is also Stalled when more vendors have said no than yes. Read the spec.

Community insights

• Sceptical

  No AI crawler uses content negotiation. Not one. They only discover the Markdown pages through the dedicated URLs, and only via the auto-discovery link.

• In favour

  Content negotiation is the standard, and it’s the one every HTTP client already knows how to speak.

• In favour

  Not every agent requests markdown today, but the ones that do are already giving their users a better experience — faster responses, lower costs, higher quality answers.

• In favour

  In the next version of Claude Code, Claude’s WebFetch tool automatically adds Accept: “text/markdown, *” to requests which helps docs sites provide token-efficient docs

Quotes are from public posts, checked against the original before they appear here. Suggest an insight, or ask for yours to be removed.