Congrats on shipping this, this looks really cool. Reminds me of n8n.io. What is the reasoning behind writing the flow editor from scratch versus utilising an existing library such as Rete.js? Also, if you need some ideas on the visual programming side of things, feel free to drop me an email, I used to work in this space.
The reason for writing a new editor is to extend the things that can be build with the Scratch/Blockly one, like frontend interfaces or, in the future, streaming flows. There were also some recurring problems people found when building with Scratch that we can solve by having total control over the editor, like reacting to type mismatches. (Note, though, that Scratch also supports type checking)
I'm really interested on other viewpoints on visual programming, so for sure I'll send that email.
We draw a lot of inspiration from Node-RED, especially for builing our own flow-based editor.
We take the same idea and try to make the first steps simpler by providing a cloud service and pre-installed connections to third party services. Also, having a collaborative editor should help when less tech-oriented people need some help to understand why some automations don't work as they expect.
There are more differences: we are working on further integrating the UI builder with the rest of the automation editor. On the other hand we don't yet support building API's directly on our flows or adding code blocks, but we are also working on that.
So, in summary, the main point is to have an easier onboarding and avoid needing to install servers and plugins from the start.
An account is needed because up until recently the only input/output possible was through connections with other services, and the way to associate user's with its services was through an account.
I know it's not ideal, but opening public programs in read-only mode is possible without an account:
https://gitlab.com/programaker-project/programaker-core
Keen to give this a go.