Content Pipeline Automation: From Idea to Published Post

Writing a daily post is easy once. It is hard to keep up for months. The part that made this sustainable for me was automating the pipeline around the writing itself. Idea capture, drafting, publishing, and distribution are all things a computer can do. Only the creative work stays with me. Here is the stack.

The Stages

The pipeline has five stages:

Capture: ideas land in a single inbox
Triage: ideas become drafts in a writing app
Draft: I write the actual content (still human)
Publish: the post goes live with correct metadata
Distribute: the post reaches newsletter, RSS, social

Only stage three is me at the keyboard. The other four are scripts.

The Idea Inbox

Every idea lives in one markdown file. No apps, no tags, no folders. When something looks like a topic I add a single line. Once a week a script promotes ideas into draft files:

bash

# triage.sh
while read -r idea; do
    slug=$(echo "$idea" | slugify)
    if [ ! -f "drafts/$slug.md" ]; then
        echo "creating draft for: $idea"
        template &lt; idea.tpl --title "$idea" > "drafts/$slug.md"
    fi
done &lt; ideas.md

That script turns a one-line idea into a pre-formatted draft with frontmatter, categories, and an outline placeholder. Takes two seconds. Removes all the activation energy.

The Publish Step

Publishing is a single make target:

plaintext

make publish SLUG=post-slug

Under the hood it validates the frontmatter, checks word count, verifies categories are real, POSTs to the API, and commits the file. If any step fails the post does not publish. That validation is what keeps the site internally consistent even when I am writing fast.

The Distribution Hook

After publish succeeds, a webhook triggers the distribution fan-out:

Add to the RSS feed (regenerated automatically)
Schedule a newsletter mention for the next digest
Post a brief link to my social account
Update the sitemap

Each of those is a separate Lambda, running async. If one fails, the others still run. The post is up either way.

What I Did Not Automate

Three things I kept manual on purpose:

Writing the actual content: this is the job, automating it defeats the point
Editing: a second pass by a human catches things the first pass misses
Cover image selection: small visual choices still benefit from a human eye

Automating the boring parts of content pipelines makes sustained publishing possible. Automating the creative parts makes the content forgettable. Know the difference.

Read the Jamstack introduction for the broader pattern.