Beyond working once.
WHAT A RELIABLE SCRIPT DOES
States what it is doing Handles failure explicitly Leaves things consistent if interrupted Can be run again safely Logs enough to diagnose
THAT FOURTH POINT
Running twice should not do the work twice, or should be harmless if it does.
WHY THAT MATTERS
Scripts get run again after an apparent failure.
WHAT TO BUILD IN
A check of preconditions before starting Clear failure, rather than partial completion A summary at the end
WHAT TO AVOID
Assuming files exist Assuming a network call succeeds Assuming data is well formed Continuing after an error
WHAT TO DO ABOUT PARTIAL FAILURE
Decide deliberately: stop entirely, or record and continue.
Then say which in the output.
WHAT TO TEST
The failure paths, not only the success path.
HOW
Give it bad input deliberately.
WHAT TO ADD BEFORE RELYING ON IT
Logging, and a way to know it ran.
