Just installed it and started using. It is timely as I have to fill a number of forms today on government sites so looking forward to using it. What are some best practices you have come up with for reducing the cognitive load for the trigger lifecycle - that is detect the need, come up with one that fits an easy to retrieve mental model? Namespacing obviously comes to mind, and some trigger design should be as conflict free as possible, yet brief:

  :i.a - address
  :i.n - name
  :i.p - phone

Debating if I should feed my zsh history to chatgpt and as it to come up with some. Any other advice from the power users?

I love espanso! The cross-platform support is huge (I use it on macOS, windows, and X11 and Wayland-based Linux systems).

Moreover, the original creator (Federico) and the current head maintainer to whom he has handed most of the day-to-day (AucaCoyan) are two of the kindest people I have ever come across in open source. All issues and contributors are treated with respect, it really is refreshing to feel so welcome when trying to contribute.

Autohotkey is go to on Windows for stuff like this, and its called hotstrings [1]. Hotstrings are much more powerfull. Trigger can for example run arbitrary function. AFAIK on linux, you can run it using wine.

[1] https://www.autohotkey.com/docs/v2/Hotstrings.htm

I found Espanso very useful, but some bugs made me move on to Raycast, BetterTouchTool, etc. for similar functionality. For example, if Espanso config file is on a cloud drive, it doesn't automatically sync or read the file upon reboot.

I'm planning to move back to Espanso though, as Raycast is moving in the wrong direction with all the AI non-features.

I think the problem Espanso is trying to solve should be addressed by GUI toolkits, like GTK, QT, etc. Otherwise, living with an authorized keylogger in our system in order to introduce unicode characters seems overkill.