A simple file-based task manager in Bash

task

#!/bin/sh

if [ -z $TASK_DIR ]; then
  TASK_DIR=$HOME/tasks
fi

help () {
  echo "Valid subcommands: create, ls, rm, cat, edit"
}

nth () {
  ls $TASK_DIR | sed "$1q;d"
}

taskcreate () {
  ESCAPE=$(printf "%q" $1)
  FILENAME="${*}.txt"
  touch "$TASK_DIR/$FILENAME"
}

taskls () {
  ls  $TASK_DIR | cat -n | sed "s/\..*//g"
}

taskrm () {
  N=$1
  if [ -z $N ] ; then
    echo "** TASKS **"
    taskls
    read -p "Select a task number: " N
  fi
  F=$(nth $N)
  rm "$TASK_DIR/$F"
}

taskcat () {
  N=$1
  if [ -z $N ] ; then
    echo "** TASKS **"
    taskls
    read -p "Select a task number: " N
  fi
  F=$(nth $N)
  cat "$TASK_DIR/$F"
}

taskedit () {
  N=$1
  if [ -z $N ] ; then
    echo "** TASKS **"
    taskls
    read -p "Select a task number: " N
  fi
  F=$(nth $N)
  $EDITOR "$TASK_DIR/$F"
}

case $1 in
  cat) taskcat $2
      ;;
  rm) taskrm $2
      ;;
  ls) taskls
      ;;
  edit) taskedit $2
      ;;
  create) taskcreate $2
      ;;
  *) help
      ;;
esac

Have you considered replacing lines 58 onward with:

case "$1" in
  cat|rm|ls|edit|create) "task$1" "$2" ;;
  *) help ;;
esac

Variables in command names are pretty universally supported by shells; I learned this dispatching trick from git-submodule.sh, and git's coding guidelines are explicitly very stringent about cross-compat.