The final Markdown

Are you, just as me, finding an increasing amount of .md (markdown) files around you? And no obvious way of opening/viewing them? Then try MDPod.

MDPod is something I created during the last weekend. This post is a bit on the "why", the "how" and the "oh, but wait, what does this mean?". And this post took longer to write than to create MDPod ("The post is written without AI").

"The why"
I have more and more markdown (.md) files on my laptop. The format has been around for quite some time, but not until the recent year have I encountered so many of them that I started to think what the best way to open/read them are. In most editors I only the got the raw format, not the user-friendly rendering.

Two of the places where .md files started to show up for me were a) on GitHub (eg most README files are in .md format and b) in interactions with LLMs / AI (both as output an input files).

"The how"
The LLM/AI/GitHub mentions above gives away how MDPod was created. Yes, it was "vibe-coded", or perhaps rather "SDD-coded" (Spec-Driven Development).

The development was a two step process. First I used ChatGPT (GPT 5.5) to get a somewhat crisp requirements and implementation specification. The prompt was: 

"I want my devs to develop a .md file viewer, potentially with basic editing functionality. The solution shall be in a single, standalone HTML file. Can you write a useful requirements/implementation specification (as a .md file, of course)" 

The resulting specification was then fed into Claude (Sonnet 5, Medium effort). Out, after only one run, came an MD viewer editor. Sort of faster than locating and downloading an MD viewer from the Internetz..

Then I went back to the ChatGPT session and discussed a suitable name. Got several suggestions, but I ended up with MDPod, which was not on the list from ChatGPT. But as soon as I asked the sycophantic behavior approved..:

After that a few short prompts to fix the naming, set the default state to "view window only", add a "help/about" button and to implement the "?file=" parameter support. But the main functionality was there after the first pass.

This was it, and I had an MD viewer/editor on my laptop. Creating a new GitHub repo and uploading the file made it possible to distribute and serve MDPod to anyone (including you), without any specific server involved (besides the GitHub infra). Low friction and low (basically zero) marginal cost and effort.

"Oh, but wait, what does this mean?"
I think "writing code" is one of the better use cases for generative AI. I suspect that many other primarily use it for other type of text and media generation (what are your primarily use case?)

The threshold for writing code is substantially lowered, and for those already writing code the time and effort needed is much shorter/smaller. Expect an abundance of code.

This in turn have several probable implications. The business model and pricing model for software will likely need to change. There will be more "lot size of one" (bespoke tools) created.

For organizations and corporations using software this sound good; "more tools, efficiency and automation at a lower cost". But this might also create "shadow IT on steroids". The cost of governance and compliance checks will not follow the same cost curve. Also, the more automation, tools and friction removal you introduce in an organization, the more you will likely the increase the overall complexity. And this comes at a cost.