Every page you publish. Readable by machines.

The plugin registers a rewrite rule that recognises URLs ending in .md. When one is requested, it resolves the path to the right post through your existing permalink structure, runs the content through the normal WordPress filters (blocks and shortcodes included), converts the resulting HTML to Markdown and serves it with the text/markdown content type. Nothing is written to disk — documents are generated on request.

The title, then metadata lines — publication date, author, categories, tags — then the converted body (headings, paragraphs, bold and italic, links, images, ordered and unordered lists, blockquotes, inline code and fenced code blocks with the language preserved), and a link back to the original page. Every metadata line is individually toggleable, and the author name can be overridden with a house byline.

No. Only published, public content is served. Drafts, pending, private and password-protected posts answer with a standard HTTP 404 — the endpoint never reveals whether restricted content exists.

A WordPress page arrives wrapped in navigation, theme markup, scripts, styling and tracking — most of it irrelevant to a system trying to read the article. Clean Markdown is the format AI assistants and LLMs parse best: the structure survives, the noise is gone, and it costs a fraction of the tokens.

Same content, two consumers. The .md version is a clean Markdown document — ideal for LLMs reading, quoting and citing. The .json version is a structured representation — title, publisher, author, ISO-8601 published and modified timestamps, canonical URL, categories, tags, excerpt, featured image and the Markdown body as a field — ideal for agents, pipelines and integrations that want data, not a document. Fields that don't exist are omitted, never invented.

Three ways. Every eligible article's HTML head carries link rel="alternate" tags (with matching Link response headers) pointing at its .md and .json twins. /llms.txt documents the URL patterns, feeds and sitemaps in one file. And /feed/ai/ is a JSON Feed 1.1 of your latest content where every item links its canonical page and both machine versions — one URL for an agent to poll.

No — that's engineered, not hoped. Every machine response carries X-Robots-Tag: noindex plus a canonical Link header pointing at the HTML article, so search engines can never index a .md or .json twin in place of your page. The HTML article keeps every ranking signal; the machine formats are for machines.

Custom post types, per-post disable controls, YAML front matter in the agent-publishing shape (publisher, ISO-8601 timestamps, category, format label, language), the structured .json endpoint, the AI feed at /feed/ai/, excerpt and featured-image metadata, HTML-table-to-Markdown conversion, an automatic output cache, a /markdown-sitemap.md index of every document, an llms.txt manifest for AI crawlers, and the option to remove the "Created by" credit line. If you also run our WP AI Visibility plugin, its llms.txt manager and the endpoint coexist cleanly — one switch decides who serves the file.

No. Documents are generated only when a machine URL is requested — normal visitors never trigger the converter. Responses are publicly cacheable with ETag and Last-Modified headers, so CDNs and repeat consumers revalidate with cheap 304s, and Pro adds a server-side cache invalidated automatically when a post or your settings change.

Visit Settings → Permalinks and press Save once. That refreshes the WordPress rewrite rules, which the endpoint needs. It requires a pretty-permalink structure (almost every site already has one).

Pay by card (Stripe) or PayPal. Your licence key is emailed after payment. Keys are domain-based — no account, no phoning home, no lock-in.

14-day money-back guarantee, no questions. Email us and we refund you.