I've been programming in Bash for decades now. I just can't imagine using it for the purposes you describe.

So many of your reasons are ridiculous, like "bash is everywhere". No, it isn't - on this new Mac I had to install it, and if I had a Windows machine, I'd have to install it.

But that isn't the reason why that's a ridiculous comment, it's because any real-world project will contain dependencies!

And here are some reasons why the whole enterprise is a poor idea.

1. Unit tests. You don't even mention them! Ever write unit tests in Bash?

I have: it's miserable and incredibly hard to do near trivial things, like "this numerical array is very close to that one".

How do you do things like "mocking out a service" in Bash? How do you even test your Bash scripts without running them against real services?

2. I am a pretty good Bash programmer, but I still struggle with some things like using associative arrays in Bash. Most people aren't anywhere near as good as I am. If I write in bash, it will be maintained by people without skill in Bash programming, which is a recipe for disaster.

3. I have all these great development tools. None of them support Bash. I have no Bash debugger, no Bash linter, no Bash packager or deployer..

4. And on the reverse side, there is no simple way for me to re-use other Bash "packages" that others have written - indeed, there is no idea of a Bash package. Suppose I, for example, needed to quickly multiply a 4x4 matrix in my program. How do I do this in bash?

5. Bash has many horrible traps that do not appear in Python, or any other language. About a decade ago, Apple released an upgrade script that forgot to put quotes around Bash variables, and a bunch of people's main drives got erased. After a lot of hard experience, I put set -euxo pipefail at the top of all my scripts, but that doesn't fix the above problem, or a lot of other problems.

6. BONUS: I write Python CLI scripts using Typer, which gives me a lovely, clean CLI API with the minimum of work by using Python's reflection and typing to deduce everything from my code.

And what do I get for giving up all of this? 60MB in an image size? In quantity, a gigabyte of hard drive is about $0.02 and a gig of memory is $2, so that's $0.12 worth of RAM, or $0.0012 worth of ROM.

Even a minute of my time is worth more than $0.12.

In conclusion, simply the issue in getting Bash programmers competent enough to write correct and productionized programs should be enough to scotch your idea, but the extra work on everyone and the lack of a module ecosystem is just as bad.

My guess is that all your scripts are nearly trivial. My other guess is that you are putting a lot of load on your coworkers that you aren't even aware of because of your unfounded preferences.