Search and AI discovery¶

Cuhlumn writes the signals that search engines and AI assistants use to find, understand, and cite a blog.

Meta tags¶

Every page has these tags:

• <title> and <meta name="description">
• <link rel="canonical"> with the absolute address on the canonical domain, except on the 404 page
• robots: index, follow, max-image-preview:large, or noindex, follow on tag pages and the 404 page
• Open Graph: og:type, og:site_name, og:title, og:description, og:url, og:image, og:image:alt
• Twitter: twitter:card with summary_large_image, twitter:title, twitter:description, twitter:image
• A link to the Atom feed

Posts also have article:published_time, article:modified_time, one article:tag for each tag, and a link to the Markdown copy.

Structured data¶

Posts include the headline, description, address, image, language, author, publisher, dates, word count, and tags. The author is a Person when the site has an author. Otherwise the author is the publisher, an Organization with the site title and logo.

The publisher gets a sameAs list with the profile addresses from the social links of the site. Email links are not in the list. A site without profile links has no sameAs list. See Appearance.

Atom feed¶

/feed.xml contains the 50 latest posts with their full HTML. Pages are not in the feed. Root-relative addresses in the content become absolute addresses.

Sitemap¶

/sitemap.xml lists the home page and every published post and page, with the update time. The home page entry is not repeated, and tag pages are not listed.

robots.txt¶

User-agent: *
Allow: /
Disallow: /api/

Sitemap: https://blog.example.com/sitemap.xml

llms.txt¶

llms.txt is an index for AI assistants:

1. The site title as a heading
2. The description as a quote
3. A line about the product, when the site has a project_url
4. A ## Posts section and a ## Pages section with links to the Markdown copies and summaries

llms-full.txt contains llms.txt and then the Markdown copy of every entry.

Markdown copies¶

/{slug}.md returns an entry as Markdown:

# How we back up SQLite

> Our nightly routine for one database file.

Published October 27, 2026 on Harbor Notes: https://blog.example.com/sqlite-backups/

We back up **one SQLite file** every night.

IndexNow¶

When a post is saved with published status and the site has indexnow set to true, the server sends the post address and the home address to https://api.indexnow.org/indexnow. Bing, Yandex, and the other IndexNow engines receive the notification. Google reads the sitemap.

• Each site has its own key, served at /{key}.txt.
• A new site gets a random key of 32 hexadecimal characters.
• A site update without indexnow_key keeps the current key.
• indexnow is on by default. Set it to false to stop the notifications of one site.
• The console shows the IndexNow switch next to the key in the Domains section.
• The key file stays available when indexnow is false.

The notification runs in the background. The API answer does not wait for it, and a failure only goes to the server log.