Morick Markdown Layer for AI

Wopisanje

AI assistants and crawlers (GPTBot, ClaudeBot, Perplexity, and others) increasingly read the web to answer users‘ questions. HTML pages built for humans are heavy and noisy for them: navigation, scripts, and styling bury the actual content.

Morick Markdown Layer for AI adds a lightweight, machine-friendly Markdown layer on top of your existing site – without changing your theme, your URLs, or anything your human visitors see.

What it does

• Markdown version of every page. Append ?ai=1 to any post, page, or archive URL and the plugin returns clean Markdown (with YAML frontmatter: title, description, canonical URL, dates, categories) instead of HTML.
• Content negotiation. Clients that send an Accept: text/markdown header receive Markdown at the normal URL – no query parameter needed.
• /llms.txt generation. The plugin dynamically serves an llms.txt index: your site name, description, and a link list of all Markdown-enabled pages, so AI agents can discover your content in one request.
• Alternate link tags. Each HTML page advertises its Markdown version via <link rel="alternate" type="text/markdown"> in the head, which is how AI crawlers actually discover and follow the Markdown layer.
• Admin checklist. A dashboard screen lists every target page with its status (Auto / Manual / Excluded), body length, sortable columns, status filters, and title search.
• Manual Markdown override. For template-driven landing pages whose HTML converts poorly, write Markdown by hand per page. Manual Markdown always wins over auto conversion.
• Automatic exclusion. Pages whose converted body falls below a configurable length threshold are excluded automatically (thin content is worse than no content for AI consumption).
• No dependencies. The HTML-to-Markdown converter is a self-contained DOM walker. No Composer packages, no external services, no API calls, no data leaves your server.

How conversion is resolved

For each Markdown request the plugin resolves in this order:

1. Manual Markdown exists for the page – serve it (threshold ignored).
2. Otherwise, if the auto-converted body meets the length threshold – serve the auto conversion.
3. Otherwise the page is excluded: no Markdown is served and no alternate link is printed.

SEO safety

The Markdown response sends a Link: <original-url>; rel="canonical" HTTP header pointing back to the HTML page, plus Cache-Control: no-store, so the Markdown layer does not compete with your HTML pages in search engines.

Performance

Converted Markdown is cached per page in transients and invalidated automatically when the post is saved. The archive (llms.txt) and page lists are capped and filterable via the mdlai_post_types and mdlai_list_limit hooks.