I have a bunch, but one that I rarely see mentioned but use all the time is memo(1) (https://github.com/aktau/dotfiles/blob/master/bin/memo).

It memoizes the command passed to it.

  $ memo curl https://some-expensive.com/api/call | jq . | awk '...'

Manually clearing it (for example if I know the underlying data has changed:

  $ memo -c curl https://some-expensive.com/api/call

In-pipeline memoization (includes the input in the hash of the lookup):

  $ cat input.txt | memo -s expensive-processor | awk '...'

This allows me to rapidly iterate on shell pipelines. The main goal is to minimize my development latency, but it also has positive effects on dependencies (avoiding redundant RPC calls). The classic way of doing this is storing something in temporary files:

  $ curl https://some-expensive.com/api/call > tmpfile
  $ cat tmpfile | jq . | awk '...'

But I find this awkward, and makes it harder than necessary to experiment with the expensive command itself.

  $ memo curl https://some-expensive.com/api/call | jq . | awk '...'
  $ memo curl --data "param1=value1" https://some-expennsive.com/api/call | jq . | awk '...'

Both of those will run curl once.

NOTE: Currently environment variables are not taken into account when hashing.