Introduction
My zsh profile is over 1000 lines at this point. A lot of that is functions I asked AI to generate for me, since it's fast, portable, and saves me a ton of typing.
Here's the thing though: the shortcuts that save me the most time aren't the clever ones. They're the dumb ones. Things like clone instead of git clone && cd, or dir instead of mkdir -p && cd. Each one only saves a second or two, but I run them so often that it adds up fast.
These are in no particular order, just the ones I reach for constantly.
Git aliases for common commands
A few one-liners I have set up as plain aliases:
1alias gcp="git cherry-pick"2alias git-append="git commit --amend --no-edit -a"
gcp is self-explanatory. git-append amends the last commit with your currently staged (and unstaged, thanks to -a) changes without touching the commit message. Great for fixing up a commit you just made before you push.
Create a branch or switch to it if it already exists
One of my most-used functions. Normally you have to remember whether a branch exists before deciding between git checkout <branch> and git checkout -b <branch>. This just does the right thing either way:
1gb() {2 if git rev-parse --verify --quiet "$1" >/dev/null; then3 git checkout "$1"4 else5 git checkout -b "$1"6 fi7}
Nuke all local changes to reset the working tree
When an experiment goes sideways or I just want to throw everything away and start clean, I run nah:
1nah() {2 git reset --hard3 git clean -df4 if [ -d ".git/rebase-apply" ] || [ -d ".git/rebase-merge" ]; then5 git rebase --abort6 fi7}
This resets tracked changes, removes untracked files and directories. No confirmation prompt, so use it carefully.
Print recent commits as ready-to-paste cherry-pick commands
Useful when you need to cherry-pick a batch of commits from one branch onto another in order:
1logs() {2 if [[ -z "$1" || "$1" =~ [^0-9] ]]; then3 echo "Usage: logs <number_of_commits>"4 return 15 fi6 git log -n "$1" --reverse --pretty=format:"gcp %h"7}
Run logs 5 and you get the last 5 commits printed oldest to newest, each one already formatted as gcp <hash> (using the alias from above). Copy, paste, done.
Create a directory and cd into it in one step
This is the one I mentioned that barely saves any time per use, but I run it constantlyn and the time save compounds:
1dir() { mkdir -p "$1" && cd "$1"; }
Make mkdir always create parent directories
I got tired of hitting "No such file or directory" errors from mkdir when the parent folder didn't exist yet, so I just overrode the default behavior globally:
1mkdir() {2 command mkdir -p "$@"3}
Now mkdir always behaves like mkdir -p.
Open nano and auto-create missing parent directories
Same idea as the mkdir override, but for opening a file in nano when the folder it lives in doesn't exist yet:
1nano() { 2 if [ $# -eq 0 ]; then 3 command nano 4 else 5 dir=$(dirname "$1") 6 if [ ! -d "$dir" ]; then 7 mkdir -p "$dir" 8 fi 9 command nano "$@"10 fi11}
Safely replace a file's contents by moving the original to trash first
Instead of overwriting a file and losing the original, replace moves it to ~/.Trash first and then opens a fresh file with the same name in nano:
1replace() { 2 if [[ $# -ne 1 ]]; then 3 echo "Usage: replace <file>" >&2 4 return 1 5 fi 6 local file="$1" 7 if [[ ! -f "$file" ]]; then 8 echo "Error: '$file' not found or not a file" >&2 9 return 110 fi11 mkdir -p ~/.Trash12 mv "$file" ~/.Trash/13 nano "$file"14}
But mainly this saves me from runing "rm ~/.Trash.
Convert HEIC images to JPG or PNG
iPhone photos default to HEIC, which isn't great for sharing or uploading. These two functions batch-convert them using ImageMagick:
1heic2jpg() { 2 if [ $# -eq 0 ]; then 3 echo "Usage: heic2jpg file1.heic [file2.heic ...]" 4 return 1 5 fi 6 for f in "$@"; do 7 if [ ! -f "$f" ]; then echo "Not found: $f"; continue; fi 8 magick "$f" "${f%.*}.jpg" 9 done10}11 12heic2png() {13 if [ $# -eq 0 ]; then14 echo "Usage: heic2png file1.heic [file2.heic ...]"15 return 116 fi17 for f in "$@"; do18 if [ ! -f "$f" ]; then echo "Not found: $f"; continue; fi19 magick "$f" "${f%.*}.png"20 done21}
Both take any number of files, so heic2jpg *.heic works fine.
Batch convert images to WebP
For getting images web-ready, this wraps img2webp with sensible defaults and skips anything that isn't actually a file:
1img2webp() { 2 if [[ $# -eq 0 ]]; then 3 echo "Usage: img2webp <file.png> [file2.png ...] or img2webp *.png" 4 return 1 5 fi 6 7 for input in "$@"; do 8 if [[ ! -f "$input" ]]; then 9 echo "Skipping: '$input' is not a file"10 continue11 fi12 local output="${input%.*}.webp"13 echo "Converting: $input → $output"14 command img2webp -lossy -q 80 "$input" -o "$output"15 done16}
Runs at quality 80, lossy, which is a good default for most web use cases. Saves a TON of storage space and bandwidth
A few misc aliases
Small ones I use daily without thinking about them:
1alias pest="./vendor/bin/pest"2alias python="python3" # For AI commands expecting it3alias py="python3" # For me when I'm lazy
The pest alias saves typing out the full vendor binary path every time I want to run tests. The python/py aliases exist because plenty of AI-generated commands assume python points to Python 3, and half the time I'm too lazy to type the 3 myself anyway.
Conclusion
None of these are groundbreaking on their own. But that's kind of the point: the small, boring shortcuts you run 50 times a day save you more time overall than the clever ones you run once a week. If you're not already keeping a running file of these for yourself, start one. Every time you catch yourself typing the same thing twice, that's a candidate.
Syntax highlighting by Torchlight.dev