GPT on Question 14, Oct 2

In Bash, use an EXIT trap and track child PIDs:

#!/usr/bin/env bash

children=()

cleanup() {
    trap - EXIT
    ((${#children[@]})) || return

    kill -TERM "${children[@]}" 2>/dev/null || true
    wait "${children[@]}" 2>/dev/null || true
}

trap cleanup EXIT
trap 'exit 129' HUP
trap 'exit 130' INT
trap 'exit 143' TERM

some_command &
children+=("$!")

another_command &
children+=("$!")

wait

This handles normal exit and catchable signals, but only guarantees cleanup of the recorded direct children.

If commands can create grandchildren, put the script in its own process group/session and terminate the whole group:

setsid ./worker.sh

Then, inside worker.sh, an EXIT trap can signal its process group:

cleanup() {
    trap - EXIT
    trap '' TERM
    kill -TERM -- "-$$" 2>/dev/null || true
}
trap cleanup EXIT

Only use kill -- -$$ when the script is definitely its own process-group leader, such as when launched with setsid; otherwise it may kill the invoking shell or unrelated processes.

No Bash trap runs after SIGKILL, kernel panic, or similar abrupt death. For a hard guarantee, use an external supervisor/cgroup—typically a systemd service with KillMode=control-group—or, on Linux, a small wrapper using prctl(PR_SET_PDEATHSIG, ...).