Concept: Create a system that can take specific markdown files created in Obsidian and process them into html files that can be uploaded to the web.

Reason: Seemed like an interesting project

Process 
1. Use a tool like Pandoc to convert `*.md` files into HTML files.
2. Put the output into a specific folder
3. Commit the changes to that folder and open a pr
4. Have a publishing system watch that repo and push the artifacts to your website whenever there is an update
5. Result. An automated blog style website that will function by continuing to take notes on projects, ideas, current progress, and anything else that is even remotely interesting

This project started today. It seemed like an interesting concept so it made sense to try and make it happen, because why not? First things first I had to solidify the general idea. I want to remove as much friction as possible between myself and publishing writing to my website. Sure there are many great tools that let you spin up a blog that does all the things, but I am not really interested in more tools. Lets just keep it simple. So it made sense to create something that would be highly automated and effectively a set it and forget it, but it should be very flexible and have the ability to be updated at any point.

The core concept centers around being able to make use of a current practice of writing notes and basic documentation about ideas for projects and interests in obsidian. These notes have been incredibly helpful in keeping track of new ideas, progress on projects and learning new things. Wouldn’t it be great if I could just have them automagically be published, if and when I wanted them to be made available?

This is where the idea was formed. Using available tools ( cronjobs, bash, git ) could I create a system that would look through my Obsidian vault for specific files, convert them to html, organize them into appropriate folders and push that to the web with no intervention on my part beyond documenting the project itself? The answer was clearly yes. The next step was to figure out how I wanted it to work.

Here is what I landed on for a first version:

  • All posts must have some basic control to figure out if it should be published or ignored.
    • Solution: Use tags. Specifically the tag ( publish )
  • All posts must be organized and timestamped
    • Solution: Use Obsidian templates to manage the creation of new notes and prefill them with timestamp properties and any other bit of useful metadata information
  • All posts must can be edited, but that must be tracked on the system as well.
    • Solution: Use the same properties concept defined above to have an edited timestamp that will update automatically and present itself on the site
    • Note: Any update or edit should not impact where the post is stored when converted to html

Future versions will have other features. Here are a few that I am currently planning

  • Automatic banner image generation based on the content of the post itself. The system should summarize the text, convert the concept into a prompt, put that prompt into an image generation pipeline, add a summary of that image to a banner image alt tag and push the resulting generated image up with the published page.
  • Table on contents for the entire project. This too should be automated and updated with each new post
  • Category filter or passive search using tags. Each post will have at least one category/tag associated with it. There should be a very simple way to select a tag and see all relevant entries related to that post.

There are more features and ideas to come, but I think for the moment this is a great start. And if I want to meet the target of creating this project within a 24 hour window ( there is no reason for this timebox other than general curiosity ) I need to move along quite quickly, but without compromising the core concept or cutting corners. It’s ok to go fast, but that is no reason to be reckless.