From Markdown to Publishing Platform: Building dout.dev's Custom CMS Without a Framework

dout.dev is a vanilla-first static site with zero runtime dependencies, but the interesting part is not only the frontend. Behind the site there is a small custom CMS that turns Markdown into a complete publishing workflow. Instead of adapting a general-purpose CMS to a narrow editorial process, the project uses a focused pipeline to parse front matter, normalize content, generate archives and taxonomies, render pages through a custom template engine, and run validation before anything is published.

This talk walks through the architecture and the tradeoffs behind that decision. I will show why a custom CMS was a better fit for this project than a larger platform, how the content model stayed intentionally small, and where custom code creates both leverage and maintenance cost. Using dout.dev as a concrete case study, I will explain how performance, accessibility, determinism, and maintainability become CMS concerns, not just frontend concerns, when the publishing pipeline is designed end to end.

Attendees will leave with a practical framework for deciding when a custom CMS is justified, what the minimum viable publishing architecture looks like, and how to keep a small content platform understandable as it grows.

What will the attendees learn from this talk

- How to decide whether a custom CMS is the right choice for a specific editorial workflow
- How to model posts, metadata, tags, archives, and series with a simple Markdown plus front matter pipeline
- How to separate content ingestion, normalization, templating, and output generation in a deterministic build flow
- How to bake accessibility, performance, and validation into the publishing system instead of treating them as frontend afterthoughts
- How to keep a small CMS maintainable with tests, conventions, and CI quality gates