Whoever is first in the morning git status typing and then drinking coffee will love this little helper. Especially when many web projects are being developed in parallel, one is worth it compact overview: Where is the working tree clean, where are unmerged changes hanging, and where is a pull/push waiting? A small shell tool will do the trick - as long as it robustly handles spaces/Unicode in paths and doesn't choke on hanging remotes.
#!/usr/bin/env bash
set -Eeuo pipefail
export LC_ALL=C
# Check if a command exists (no output).
have() { command -v "$1" >/dev/null 2>&1; }
# Ensure we have a `sort` that supports -z (NUL-delimited) input.
SORT_BIN="sort"
if ! "$SORT_BIN" -z </dev/null 2>/dev/null; then
if have gsort && gsort -z </dev/null 2>/dev/null; then
SORT_BIN="gsort"
else
printf 'Error: This script requires "sort -z" (GNU coreutils). Install coreutils (gsort).\n' >&2
exit 1
fi
fi
# Use GNU `timeout` if available; otherwise try `gtimeout` (macOS); otherwise no timeout.
TIMEOUT_BIN="timeout"
if ! have "$TIMEOUT_BIN"; then
if have gtimeout; then
TIMEOUT_BIN="gtimeout"
else
TIMEOUT_BIN=""
fi
fi
# Require git.
if ! have git; then
printf 'Error: "git" not found.\n' >&2
exit 1
fi
# Remove a leading "./" from a path for cleaner output.
trim_dot_slash() {
case "$1" in
./*) printf '%s\n' "${1#./}" ;;
*) printf '%s\n' "$1" ;;
esac
}
# Legend + divider (as requested)
printf '\n🟢: clean\n🟡: behind/ahead\n🔴: modified\n\n----------------------------------\n\n'
# Find all .git directories, NUL-delimited; sort NUL-delimited; iterate safely.
find . -type d -name .git -print0 \
| "$SORT_BIN" -z \
| while IFS= read -r -d '' gitdir; do
repo="${gitdir%/.git}"
display_path="$(trim_dot_slash "$repo")"
# Skip anything that isn't a proper work tree (safety check).
if ! git -C "$repo" rev-parse --is-inside-work-tree >/dev/null 2>&1; then
continue
fi
# Working tree status; include untracked files for a strict "red" signal.
status_out="$(git -C "$repo" status --porcelain=v1 || true)"
# Upstream divergence check (only if an upstream is configured).
ahead=0
behind=0
if git -C "$repo" rev-parse --abbrev-ref --symbolic-full-name '@{u}' >/dev/null 2>&1; then
# Refresh refs; protect with timeout so a hanging remote doesn't stall the loop.
if [ -n "$TIMEOUT_BIN" ]; then
"$TIMEOUT_BIN" 10s git -C "$repo" fetch --all --prune >/dev/null 2>&1 || true
else
git -C "$repo" fetch --all --prune >/dev/null 2>&1 || true
fi
# Count commits only on our side (ahead) and only on upstream's side (behind).
ahead="$(git -C "$repo" rev-list --count --left-only HEAD...@{u} 2>/dev/null || echo 0)"
behind="$(git -C "$repo" rev-list --count --right-only HEAD...@{u} 2>/dev/null || echo 0)"
fi
# Decide the signal:
# - RED if the working tree isn't clean
# - YELLOW if clean but ahead/behind of upstream
# - GREEN otherwise
if [ -n "$status_out" ]; then
printf '🔴 %s\n' "$display_path"
else
if [ "${ahead:-0}" -gt 0 ] || [ "${behind:-0}" -gt 0 ]; then
printf '🟡 %s\n' "$display_path"
else
printf '🟢 %s\n' "$display_path"
fi
fi
doneThe script still needs to be... chmod +x ~/path/to/script.sh make it executable and can set up an alias to save valuable typing: Here you add to his ~/.bashrc / ~/.zshrc / ~/.bash_profile the entry alias gscan='bash /path/to/script.sh' in addition. From then on, a simple gscan in the desired root directory.
A word about why the second run is noticeably faster: During the first run, the file system still has to scan through everything; after that is metadata and many .git-Structures have landed in the operating system's page cache. References and commit graphs have also already been warmed up. The next fetch It now mostly only transmits small deltas. No dashboard, no overhead – a quick status snapshot directly in the terminal.