GuideRouting and navigation

Feeds, sitemaps and other outputs

Generate RSS feeds, sitemaps, llms.txt, robots.txt and JSON files from templates.

Not every file of a site is a page. Feeds, sitemaps, robots.txt, llms.txt and JSON files are generated from the site's data at build time. In Douze they are templates in the content folder: feed.xml.liquid is rendered like a page and written as feed.xml. This page shows the common ones and the filters they use.

How output templates workLink to this heading

A file named name.ext.liquid in the content folder:

  • has front matter, like a page, and is rendered as Liquid with the page's data, site, site data and collections;
  • is written to name.ext, or to its permalink;
  • is not in the collections, and gets no layout: layouts only apply to HTML files.

.txt files of the content folder are rendered as Liquid too, so robots.txt needs no .liquid extension. .json, .xml, .yml and .yaml files are data files of their folder and are not published: write api.json.liquid or feed.xml.liquid, not api.json or feed.xml.

RSS feedLink to this heading

_content/feed.xml.liquid
<?xml version="1.0" encoding="utf-8"?>
<rss version="2.0">
<channel>
  <title>My blog</title>
  <link>{{ "/" | absolute_url }}</link>
  <description>Posts of my blog</description>
  {%- assign posts = collections.post | sort_by: "data.date", "desc" %}
  {%- for post in posts limit: 20 %}
  <item>
    <title>{{ post.data.title }}</title>
    <link>{{ post.data.url | replace: 'index.html', '' | absolute_url }}</link>
    <guid>{{ post.data.url | replace: 'index.html', '' | absolute_url }}</guid>
    <pubDate>{{ post.data.date | date_to_rfc822 }}</pubDate>
    <description>{{ post.data.description }}</description>
  </item>
  {%- endfor %}
</channel>
</rss>
  • collections.post holds the pages tagged post, sorted by date ascending; sort_by: "data.date", "desc" puts the newest first.
  • absolute_url prepends site.url, set in douze.yml.
  • A date without a time in front matter (date: 2026-01-02) is midnight in the time zone of the machine that builds, so the feed differs between machines. Write dates with a time zone (date: 2026-01-02T00:00:00Z) for the same output everywhere.

The items of collections have the data of the pages, not their content. For full content in a feed, add a second collect step with content: true. See Add an RSS feed and the collect step.

SitemapLink to this heading

_content/sitemap.xml.liquid
<?xml version="1.0" encoding="utf-8"?>
<urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9">
{%- for p in collections.all %}
  <url><loc>{{ p.data.url | replace: 'index.html', '' | absolute_url }}</loc><lastmod>{{ p.data.page.date | date_to_xmlschema }}</lastmod></url>
{%- endfor %}
</urlset>

page.date is the file's modification time, and date the front matter date when the page has one. On a multilingual site, collections.all | locale_pages lists the pages of every locale, fallback copies included. See Add a sitemap.

robots.txtLink to this heading

_content/robots.txt
User-agent: *
Sitemap: {{ "/sitemap.xml" | absolute_url }}

llms.txtLink to this heading

_content/llms.txt.liquid
# My site

> Documentation of My product.

{% for p in collections.all %}
- [{{ p.data.title }}]({{ p.data.url | replace: 'index.html', '' | absolute_url }}): {{ p.data.description }}
{%- endfor %}

See Build with AI for Markdown versions of every page.

JSONLink to this heading

_content/posts.json.liquid
---
permalink: /api/posts.json
---
[
{%- for post in collections.post %}
  { "title": {{ post.data.title | json }}, "url": "{{ post.data.url | replace: 'index.html', '' }}" }{% unless forloop.last %},{% endunless %}
{%- endfor %}
]

json (or jsonify) writes a value as JSON, with its quotes and escapes. A whole object or list works too: {{ site.authors | json }}. Inside quotes, {{ }} writes the content of a JSON string: "{{ post.data.url }}".

For many JSON files generated from data, rather than one template per file, use a liquid step with a template whose permalink ends in .json: its outputs are encoded for JSON too. See Pages from data and the liquid step.

Filters for outputsLink to this heading

Filter Result
absolute_url site.url followed by the url
relative_url The url after the path of site.url (/blog/x/ for https://example.com/blog)
xml_escape &, <, >, " and ' as entities, for a step with encode: none
strip_html The text without tags, comments, scripts and styles
date_to_xmlschema 2020-01-02T03:04:05+00:00
date_to_rfc822 Thu, 02 Jan 2020 03:04:05 +0000
sort_by: "data.date", "desc" The items sorted by the value at a path, missing values last
where_exp: "item", "item.data.draft != true" The items for which the expression is true
json, jsonify The value as JSON
excerpt The first paragraph of HTML, or what comes before <!-- more -->

See Liquid filters and tags for every filter.

See alsoLink to this heading

Esc to close