It's one of my favourite tools, the entire Picolisp system is like 1,5 MB or so, so it's really easy to move to some constrained or remote environment, as long as it's POSIX.

When terminal incantations grow unwieldy I usually abstract by put them in a Picolisp function and start doing things in the REPL. It takes like ten minutes to write a wrapper around a MySQL CLI client that takes a query string and outputs a list of lists, and then you can wrap that in fork/bye and mapcar it on cuts from a list to run queries or imports in parallel. Similarly you can trivially hack up an 'Ansible at home' by running commands over SSH in parallel. If I'm worried about data races I let Linux handle it by writing lines to a file and then slurp that back in when every process is done.

Surprisingly many things are heavier or slower than launching and disbanding a POSIX process, so it's often quite viable to just hand over to forks. Once I scraped out data about many thousands of people in tens of organisations by wrapping w3m and forking on URL:s I'd gathered in a similar way. Can probably be done with Beautiful Soup too, but I already knew how to use a browser to grab a copy of a web page and low level string parsing was probably faster and easier to implement than some abstracted XML representation or XPath. The value of that data set was easily a couple of orders of magnitude larger than what I got payed to assemble it.

I mean, you can do these things in bash or Python or Clojure or something, but the Picolisp REPL has really good ergonomics and gets out of the way. Performance isn't great, but I find it good enough and have handled files with hundreds of thousands of lines without getting annoyed. Sometimes I reach for Elixir instead but it feels a bit clumsy in comparison.