Using draft posts to control content on this blog

Jun 17, 2024

On this blog, I have the ability to create drafts that allow me to work on multiple articles in parallel and also control the release schedule for my content.

The Requirements

To start with, I laid out what I wanted from a draft system:

  1. An article marked as draft should not be listed on the blog at all.
  2. If I set the publication date of an article in the future, it should not be listed on the blog until said date.
  3. In development mode, I want all articles to be listed, with status badges for "draft" and "ready to publish" articles.
  4. I want to avoid manual builds of my blog just to list "ready to go" articles. This is a requirement as this blog is statically built.

The Solution

Once I had the requirements finalized, I set out to create my system. For drafts, I used 11ty's custom collections to create a collection that filters out any article that doesn't have the published flag, but only when I'm building a site for deploying. When in development mode, I want to see all articles.

// in .eleventy.config.js
eleventyConfig.addCollection("articles", function (collectionApi) {
  return collectionApi.getFilteredByTag("articles")
    .filter(function (article) {
      if (process.env.BUILD_DRAFTS === "true") return true;
      return "published" in &&;
    .sort(function (a, b) {
      return -;

// server config
if (process.env.ELEVENTY_RUN_MODE === "serve") {
  process.env.BUILD_DRAFTS = "true";

For the publication date, I used the date frontmatter property. I implemented a check comparing the article's date against the current date using luxon's helper methods. 11ty uses UTC to parse any dates and times, so I parse my article's date as UTC to avoid off-by-one errors.

// the rest of the config file...

config.addCollection("articles", function (collectionApi) {
  return collectionApi.getFilteredByTag("articles")
    .filter(function (article) {
      if (process.env.BUILD_DRAFTS === "true") return true;
      const now = DateTime.utc();
      const articleDate = DateTime.fromJSDate(, {zone: "UTC"});
      return "published" in && && now >= articleDate;
    .sort(function (a, b) {
      return -;

In order to get the status badges for articles, I used Nunjucks's macro tag to create a function that displays a <small> tag that says either DRAFT or READY depending on each article's date and published frontmatter properties.

{% macro articles(list) %}
<div class="grid grid-cols-1 sm:grid-cols-2 grid-flow-row items-stretch gap-4">
  {%- for post in list -%}
  <article class="mt-2 mb-4 [&_a_h3]:hover:underline flex flex-col justify-between gap-4">
    <a href="{{ post.url }}" class="no-underline">
      <h3 class="mb-4">{{ }}</h3>
      <p class="font-light tracking-wider">{{ }}</p>
    <div class="flex justify-between items-center">
      <small class="uppercase">{{ | readableDate }}</small>
      {% if not %}
      <small class="bg-amber-100 text-amber-600 py-2 px-3 rounded-full uppercase">&#9888; draft</small>
      {% endif %}
      {% if | isDateInFuture and %}
      <small class="bg-blue-200 text-blue-600 py-2 px-3 rounded-full uppercase">&#x2714; ready</small>
      {% endif %}
  {%- endfor -%}
{% endmacro %}

And here's how that looks when I'm running the site in development mode (along with a sneak peek at upcoming articles):

Screenshot of a grid of article titles

That's all I could do within 11ty. To get the automated deployments going, I needed to go to the deployment flow. My blog is deployed on Azure Static Web Apps using a GitHub Actions script that Azure generated for me. This script will build and deploy the site any time a commit is made to the main branch of the git repository, either through a direct commit or a PR getting merged into main. This setup is already partially automated, but I'd still only get new posts when I actually made a commit. The limit to this approach is that if I have an article "ready-to-go", i.e. I'm finished with it but want to publish it in the future, I still have to either do a manual deployment on the date I want the article to go live on the blog, or push a (probably) unrelated commit. Instead, I want the site to deploy without any manual intervention.

Lucky for me, GitHub Actions supports running jobs on a schedule using cron syntax. The full script is on the GitHub repo here. Most of it comes auto-generated from Azure, with my changes being limited to setting up the schedule itself and updating the execution check on the build_and_deploy_job job. With this, my blog's deployment is fully automated to my satisfaction, and the only thing I now need to do is write the content and make sure any articles I want to see on the site have a valid date and the published flag on their frontmatter.

The implementation isn't perfect, there's a lot of duplicate code that I can clean up, along with some other minor improvements, like process.env.BUILD_DRAFTS being set to the string "true". I plan to keep refining it to the point where I'm satisfied with it, but it's good enough right now for a v1. After all, perfect is the enemy of good.

If you have any questions or feedback about this article, feel free to email me at feedback (@) thesilverhand (.) blog.

← Next Article

My new favourite keyboard