Azey

AI agents from ChatGPT, Claude, Perplexity, and others read your content to generate answers — but they parse through HTML, JavaScript, and navigation just to reach it. Serve Markdown gives them a direct path: clean Markdown with structured metadata. When a crawler requests it — via an <code>Accept: text/markdown</code> header or a <code>.md</code> URL — your site responds instantly. No noise, no guesswork.
<h4>How It Works</h4>
Serve Markdown adds three capabilities to your WordPress site:
1. Content Negotiation 
When an AI crawler sends <code>Accept: text/markdown</code>, your site returns Markdown instead of HTML — no URL changes needed.
2. .md URL Suffix 
Append <code>.md</code> to any post or page URL (e.g. <code>example.com/my-article.md</code>) to get the Markdown version directly.
3. Markdown Auto-Discovery 
Every page includes a <code><link rel="alternate" type="text/markdown"></code> tag in the HTML head so crawlers can find your Markdown automatically.
<h4>What the Markdown Output Looks Like</h4>
Every Markdown response includes YAML frontmatter with structured metadata followed by your post content. Here is a real example:
<pre>
---
url: 'https://akumarjain.com/textexpander-year-in-review-2025/'
title: TextExpander Year in Review 2025
author:
 name: Ajay
 url: 'https://akumarjain.com/author/akumarjain/'
date: '2026-01-05T09:05:00+05:30'
modified: '2026-01-10T14:30:06+05:30'
type: post
categories:
 - 2025
 - Year in Review
tags:
 - '#ToolsIUse'
 - Productivity
 - Remote Work
image: 'https://akumarjain.com/wp-content/uploads/text-expander-year-in-review-2025.webp'
published: true
---

# TextExpander Year in Review 2025

In my daily work, I need to type the same words and phrases repeatedly...
</pre>
<h4>Features</h4>
Content Serving
<ul>
<li>Content negotiation via <code>Accept: text/markdown</code> header</li>
<li><code>.md</code> URL suffix on any post or page</li>
<li>Auto-discovery <code><link rel="alternate"></code> tag in page head</li>
<li>Each feature can be toggled independently</li>
<li>Choose which post types are exposed (posts, pages, custom post types)</li>
</ul>
Frontmatter and Metadata
<ul>
<li>YAML frontmatter with title, author, date, categories, tags, featured image, and more</li>
<li>Toggle individual metadata fields on or off</li>
<li>Add custom static key-value pairs (e.g., <code>license</code>, <code>language</code>)</li>
<li>Map WordPress custom fields (post meta) into the frontmatter</li>
</ul>
Access Control
<ul>
<li>Per-post opt-out via a sidebar checkbox in the editor</li>
<li>Exclude entire categories from Markdown serving</li>
<li>Exclude entire tags from Markdown serving</li>
<li>Only published posts are served — drafts and private posts are never exposed</li>
</ul>
Crawler Insights
<ul>
<li>Request log showing every Markdown request with timestamp, URL, bot name, and method</li>
<li>Automatic bot detection: ClaudeBot, GPTBot, ChatGPT, OAI-SearchBot, PerplexityBot, Googlebot, Bingbot, Cohere, Meta AI, Bytespider, Applebot, and more</li>
<li>Stats dashboard: total requests, requests today, unique bots</li>
<li>Filter log by individual bot</li>
<li>Configurable retention period with automatic pruning</li>
<li>One-click log clearing</li>
</ul>
Editor Integration
<ul>
<li>“Preview Markdown” button in the post editor sidebar</li>
<li>“View Markdown” link in the Posts and Pages list table</li>
<li>Meta box shows on all enabled post types</li>
</ul>
<h4>Performance</h4>
Markdown conversion only runs when specifically requested — regular visitors see the same HTML pages they always have. The crawler log uses a lightweight custom table with three safeguards against unbounded growth: time-based retention, a row count cap, and a table size cap.
<h4>Privacy</h4>
The crawler log stores IP addresses and user-agent strings from Markdown requests. This data stays in your WordPress database and is never sent to external services. You can disable logging entirely, configure a retention period, or clear the log at any time from the settings page.

Serve Markdown

Serve Markdown versions of your posts and pages to AI agents and crawlers. Content negotiation, .md URLs, auto-discovery, and crawler logging.

Serkan Algur

Markdown Content Negotiator for LLMs is a performance-optimized WordPress plugin designed to serve your website’s content in a format that AI agents and LLMs (Large Language Models) love: Clean Markdown.
Using standard HTTP Content Negotiation, the plugin detects when a request is made with the <code>Accept: text/markdown</code> header. Instead of serving the standard HTML theme, it returns a Markdown version of the post or page, complete with YAML Frontmatter, token estimation headers, and customizable content signals.
To ensure maximum performance and minimal server load, Markdown versions are pre-generated and cached using WP-Cron.
<h4>Key Features</h4>
<ul>
<li>Content Negotiation: Automatically switches to Markdown output when requested via the <code>Accept: text/markdown</code> header.</li>
<li>YAML Frontmatter: Includes metadata like Title, Date, Author, and Categories in a structured format.</li>
<li>Performance Tracking: Provides an <code>X-Markdown-Tokens</code> header using standard LLM token heuristics.</li>
<li>AI Metadata: Includes <code>X-Content-Signal</code> headers to help agents understand the nature of the document.</li>
<li>Background Caching: Uses WP-Cron to pre-calculate Markdown strings, ensuring zero latency during requests.</li>
<li>Admin Settings: Choose which post types to enable and configure global AI content signals.</li>
</ul>

Markdown Content Negotiator for LLMs

Detects Accept: text/markdown and serves pre-generated Markdown versions of posts and pages for AI agents and LLMs.

Serve Markdown vs Markdown Content Negotiator for LLMs