Hi PH 👋 Olivier, solo maker of an open-source C4 architecture tool
Hi everyone! I'm Olivier, a software architect from Canada, building MapleSpire on my own.
The problem I kept hitting: architecture diagrams rot. They live in slides, wikis and whiteboard photos, each one a slightly different copy of the truth, and nobody knows which one is current.
Good tools for this exist, but the ones that did what I needed were priced per seat, well beyond what I could pay myself, and beyond what the company I work for was willing to spend. And I couldn't find an open-source one that covered the whole job: a live, collaborative model, not just diagrams-as-code. So I decided to build the first version that meets my own needs, and ship it.
It treats architecture as a model, not a drawing (C4: people, systems, containers, components). Every diagram is a view of that one model, so renaming a service once updates every diagram, table and export. It works offline, syncs when you're back, and it's free and open source (AGPL-3.0).
I'm launching here in a few weeks and I'd love to learn from this community first:
1. If you document architecture today, what makes your diagrams go stale?
2. Anything you wish you'd known before your first launch?
Happy to return the favour: if you're building dev tools, drop a link and I'll give honest feedback.
Replies
Welcome Olivier! Honest answer to question 1: I stopped drawing diagrams. My architecture lives as text in the notes file my coding agents read, a folder tree with one line per part and the rules about which layer may depend on which. It stays fairly current for one reason: the agent reads it on every task, so a wrong line gets noticed fast. A drawing nobody consults goes stale because nothing ever reads it. So one idea for MapleSpire: if the model could be exported as plain text that sits in the repo, agents would read it too, and you'd get a reader that checks it every day.