A knowledge base that saves itself
The one thing standing between the reader and a product a client can maintain alone, with the approach that has already been proven elsewhere and the costs it carries
The problem
Adding an entry to a knowledge base currently means putting a file into a folder and running one command. That is fine for a developer and a wall for everyone else. It is stated plainly on the Tools You Can Own page rather than hidden, because discovering it after paying would be worse. But it is the single thing keeping this from being a product a client maintains without help.
The approach
Make the knowledge base a single HTML file that saves a new copy of itself. The client opens the file, clicks edit, types, clicks save, and the browser hands them an updated file. No Python, no terminal, no hosting, no account.
This is not speculative. TiddlyWiki has worked this way since 2004 and non-technical people have used it successfully for two decades. It is a proven pattern, not an experiment.
It also fits what already exists here. sskb.html is already a single self-contained file of about 400KB with its content baked in as data. The architecture is most of the way there.
What it would take
- Content moves out of the Python build and into a data structure inside the file
- An edit mode: add, change and remove entries
- A save function that rewrites the file's own source with the updated data and triggers a download
- On Chromium browsers, the File System Access API allows saving in place instead, with download as the fallback for Safari and Firefox
The costs, stated honestly
- Version control falls to the client. They end up with
knowledge-base.html,knowledge-base (1).html, and so on. This is the thing TiddlyWiki users most often struggle with. - Images need a policy. Embedding them directly makes the file grow quickly. Linking to them means the file is no longer self-contained.
- No two people can edit at once, and there is no way to merge two versions.
- File size grows with content, though 400KB for 113 cards suggests there is plenty of headroom.
The recommendation
Split it. Keep the Python builder for the initial build, because it is the right tool for migrating forty entries at once. Ship the client the self-saving editor for adding entries afterward. You get good tooling for the hard part, they never see a command line, and it is plausibly the difference between a low and a high price for the same offer.
Decision needed
This is a real piece of work, not an afternoon. It is worth deciding whether it belongs before or after the first sale. Before means a stronger offer and a longer wait. After means learning what clients actually ask for first, at the cost of doing some of their content updates yourself in the meantime.