home / skills / thebushidocollective / han / shell-best-practices

This skill helps you write robust, portable shell scripts by applying modern best practices for error handling, security, and readability.

npx playbooks add skill thebushidocollective/han --skill shell-best-practices

Review the files below or copy the command above to add this skill to your agents.

Files (1)
SKILL.md
11.4 KB
---
name: shell-best-practices
user-invocable: false
description: Use when writing shell scripts following modern best practices. Covers portable scripting, Bash patterns, error handling, and secure coding.
allowed-tools:
  - Read
  - Write
  - Edit
  - Bash
  - Grep
  - Glob
---

# Shell Scripting Best Practices

Comprehensive guide to writing robust, maintainable, and secure shell scripts following modern best practices.

## Script Foundation

### Shebang Selection

Choose the appropriate shebang for your needs:

```bash
# Portable bash (recommended)
#!/usr/bin/env bash

# Direct bash path (faster, less portable)
#!/bin/bash

# POSIX-compliant shell (most portable)
#!/bin/sh

# Specific shell version
#!/usr/bin/env bash
# Requires Bash 4.0+
```

### Strict Mode

Always enable strict error handling:

```bash
#!/usr/bin/env bash
set -euo pipefail

# What these do:
# -e: Exit immediately on command failure
# -u: Treat unset variables as errors
# -o pipefail: Pipeline fails if any command fails
```

For debugging, add:

```bash
set -x  # Print commands as they execute
```

### Script Header Template

```bash
#!/usr/bin/env bash
set -euo pipefail

# Script: script-name.sh
# Description: Brief description of what this script does
# Usage: ./script-name.sh [options] <arguments>

readonly SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
readonly SCRIPT_NAME="$(basename "${BASH_SOURCE[0]}")"
```

## Variable Handling

### Always Quote Variables

Prevents word splitting and glob expansion:

```bash
# Good
echo "$variable"
cp "$source" "$destination"
if [ -f "$file" ]; then

# Bad - can break on spaces/special chars
echo $variable
cp $source $destination
if [ -f $file ]; then
```

### Use Meaningful Names

```bash
# Good
readonly config_file="/etc/app/config.yml"
local user_input="$1"
declare -a log_files=()

# Bad
readonly f="/etc/app/config.yml"
local x="$1"
declare -a arr=()
```

### Default Values

```bash
# Use default if unset
name="${NAME:-default_value}"

# Use default if unset or empty
name="${NAME:-}"

# Assign default if unset
: "${NAME:=default_value}"

# Error if unset (with message)
: "${REQUIRED_VAR:?Error: REQUIRED_VAR must be set}"
```

### Readonly and Local

```bash
# Constants
readonly MAX_RETRIES=3
readonly CONFIG_DIR="/etc/myapp"

# Function-local variables
my_function() {
    local input="$1"
    local result=""
    # ...
}
```

## Error Handling

### Exit Codes

Use meaningful exit codes:

```bash
# Standard codes
readonly EXIT_SUCCESS=0
readonly EXIT_FAILURE=1
readonly EXIT_INVALID_ARGS=2
readonly EXIT_NOT_FOUND=3

# Exit with code
exit "$EXIT_FAILURE"
```

### Trap for Cleanup

```bash
cleanup() {
    local exit_code=$?
    # Clean up temporary files
    rm -f "${temp_file:-}"
    # Restore state if needed
    exit "$exit_code"
}

trap cleanup EXIT

# Script continues...
temp_file=$(mktemp)
```

### Error Messages

```bash
error() {
    echo "ERROR: $*" >&2
}

warn() {
    echo "WARNING: $*" >&2
}

die() {
    error "$@"
    exit 1
}

# Usage
[[ -f "$config_file" ]] || die "Config file not found: $config_file"
```

### Validate Inputs

```bash
validate_args() {
    if [[ $# -lt 1 ]]; then
        die "Usage: $SCRIPT_NAME <input_file>"
    fi

    local input_file="$1"
    [[ -f "$input_file" ]] || die "File not found: $input_file"
    [[ -r "$input_file" ]] || die "File not readable: $input_file"
}
```

## Functions

### Function Definition

```bash
# Document functions
# Process a log file and extract errors
# Arguments:
#   $1 - Path to log file
#   $2 - Output directory (optional, default: ./output)
# Returns:
#   0 on success, 1 on failure
process_log() {
    local log_file="$1"
    local output_dir="${2:-./output}"

    [[ -f "$log_file" ]] || return 1

    grep -i "error" "$log_file" > "$output_dir/errors.log"
}
```

### Return Values

```bash
# Return status
is_valid() {
    [[ -n "$1" && "$1" =~ ^[0-9]+$ ]]
}

if is_valid "$input"; then
    echo "Valid"
fi

# Capture output
get_config_value() {
    local key="$1"
    grep "^${key}=" "$config_file" | cut -d= -f2
}

value=$(get_config_value "database_host")
```

## Conditionals

### Use [[ ]] for Tests

```bash
# Good - [[ ]] is more powerful and safer
if [[ -f "$file" ]]; then
if [[ "$string" == "value" ]]; then
if [[ "$string" =~ ^[0-9]+$ ]]; then

# Avoid - [ ] has limitations
if [ -f "$file" ]; then
if [ "$string" = "value" ]; then
```

### Numeric Comparisons

```bash
# Use (( )) for arithmetic
if (( count > 10 )); then
if (( a == b )); then
if (( x >= 0 && x <= 100 )); then

# Or -eq/-lt/-gt in [[ ]]
if [[ "$count" -gt 10 ]]; then
```

### String Comparisons

```bash
# Equality
if [[ "$str" == "value" ]]; then

# Pattern matching
if [[ "$str" == *.txt ]]; then

# Regex matching
if [[ "$str" =~ ^[a-z]+$ ]]; then

# Empty/non-empty
if [[ -z "$str" ]]; then  # empty
if [[ -n "$str" ]]; then  # non-empty
```

## Loops

### Iterate Over Files

```bash
# Good - handles spaces in filenames
for file in *.txt; do
    [[ -e "$file" ]] || continue  # Skip if no matches
    process "$file"
done

# With find for recursive
while IFS= read -r -d '' file; do
    process "$file"
done < <(find . -name "*.txt" -print0)

# Bad - breaks on spaces
for file in $(ls *.txt); do  # Don't do this
```

### Read Lines from File

```bash
# Correct - preserves whitespace
while IFS= read -r line; do
    echo "$line"
done < "$filename"

# With process substitution
while IFS= read -r line; do
    echo "$line"
done < <(some_command)
```

### Iterate with Index

```bash
files=("one.txt" "two.txt" "three.txt")

for i in "${!files[@]}"; do
    echo "Index $i: ${files[i]}"
done
```

## Arrays

### Declaration and Usage

```bash
# Indexed array
declare -a files=()
files+=("file1.txt")
files+=("file2.txt")

# Access all elements
for f in "${files[@]}"; do
    echo "$f"
done

# Array length
echo "${#files[@]}"

# Associative array (Bash 4+)
declare -A config
config[host]="localhost"
config[port]="8080"

echo "${config[host]}"
```

### Array Best Practices

```bash
# Quote expansions
"${array[@]}"   # All elements, word-split
"${array[*]}"   # All elements, single string

# Check if empty
if [[ ${#array[@]} -eq 0 ]]; then
    echo "Empty array"
fi

# Check for key (associative)
if [[ -v config[key] ]]; then
    echo "Key exists"
fi
```

## Command Execution

### Check Command Existence

```bash
# Preferred method
if command -v docker &>/dev/null; then
    echo "Docker is installed"
fi

# In conditionals
require_command() {
    command -v "$1" &>/dev/null || die "Required command not found: $1"
}

require_command git
require_command docker
```

### Capture Output and Status

```bash
# Capture output
output=$(some_command)

# Capture output and status
if output=$(some_command 2>&1); then
    echo "Success: $output"
else
    echo "Failed: $output" >&2
fi

# Check status without output
if some_command &>/dev/null; then
    echo "Command succeeded"
fi
```

### Safe Command Substitution

```bash
# Use $() not backticks
result=$(command)      # Good
result=`command`       # Avoid

# Nested substitution
result=$(echo $(date)) # Works with $()
```

## Portability

### POSIX vs Bash

| Feature | POSIX | Bash |
|---------|-------|------|
| Test syntax | `[ ]` | `[[ ]]` |
| Arrays | No | Yes |
| `$()` | Yes | Yes |
| `${var//pat/rep}` | No | Yes |
| `[[ =~ ]]` regex | No | Yes |
| `(( ))` arithmetic | No | Yes |

### Portable Alternatives

```bash
# Instead of [[ ]], use [ ] with quotes
if [ -f "$file" ]; then
if [ "$str" = "value" ]; then

# Instead of (( )), use [ ] with -eq
if [ "$count" -gt 10 ]; then

# Instead of ${var//pat/rep}
echo "$var" | sed 's/pat/rep/g'

# Instead of arrays, use space-separated strings
files="one.txt two.txt three.txt"
for f in $files; do
    echo "$f"
done
```

## Security

### Avoid Eval

```bash
# Bad - code injection risk
eval "$user_input"

# Better - use arrays for command building
cmd=("grep" "-r" "$pattern" "$directory")
"${cmd[@]}"
```

### Sanitize Inputs

```bash
# Validate expected format
if [[ ! "$input" =~ ^[a-zA-Z0-9_-]+$ ]]; then
    die "Invalid input format"
fi

# Escape for use in commands
escaped=$(printf '%q' "$input")
```

### Temporary Files

```bash
# Secure temp file creation
temp_file=$(mktemp) || die "Failed to create temp file"
trap 'rm -f "$temp_file"' EXIT

# Secure temp directory
temp_dir=$(mktemp -d) || die "Failed to create temp dir"
trap 'rm -rf "$temp_dir"' EXIT
```

## Logging

### Basic Logging

```bash
readonly LOG_FILE="/var/log/myapp.log"

log() {
    local level="$1"
    shift
    echo "[$(date '+%Y-%m-%d %H:%M:%S')] [$level] $*" | tee -a "$LOG_FILE"
}

log_info() { log "INFO" "$@"; }
log_warn() { log "WARN" "$@" >&2; }
log_error() { log "ERROR" "$@" >&2; }

# Usage
log_info "Starting process"
log_error "Failed to connect"
```

### Verbose Mode

```bash
VERBOSE="${VERBOSE:-false}"

debug() {
    if [[ "$VERBOSE" == "true" ]]; then
        echo "DEBUG: $*" >&2
    fi
}

# Enable with: VERBOSE=true ./script.sh
```

## Complete Script Template

```bash
#!/usr/bin/env bash
set -euo pipefail

# =============================================================================
# Script: example.sh
# Description: Template demonstrating shell best practices
# Usage: ./example.sh [options] <input_file>
# =============================================================================

readonly SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
readonly SCRIPT_NAME="$(basename "${BASH_SOURCE[0]}")"

# Exit codes
readonly EXIT_SUCCESS=0
readonly EXIT_FAILURE=1
readonly EXIT_INVALID_ARGS=2

# Logging functions
log_info() { echo "[INFO] $*"; }
log_error() { echo "[ERROR] $*" >&2; }

# Error handling
die() {
    log_error "$@"
    exit "$EXIT_FAILURE"
}

cleanup() {
    local exit_code=$?
    rm -f "${temp_file:-}"
    exit "$exit_code"
}
trap cleanup EXIT

# Argument parsing
usage() {
    cat <<EOF
Usage: $SCRIPT_NAME [options] <input_file>

Options:
    -h, --help      Show this help message
    -v, --verbose   Enable verbose output
    -o, --output    Output directory (default: ./output)

Examples:
    $SCRIPT_NAME input.txt
    $SCRIPT_NAME -v -o /tmp/output input.txt
EOF
}

parse_args() {
    local OPTIND opt
    while getopts ":hvo:-:" opt; do
        case "$opt" in
            h) usage; exit "$EXIT_SUCCESS" ;;
            v) VERBOSE=true ;;
            o) OUTPUT_DIR="$OPTARG" ;;
            -) case "$OPTARG" in
                   help) usage; exit "$EXIT_SUCCESS" ;;
                   verbose) VERBOSE=true ;;
                   output=*) OUTPUT_DIR="${OPTARG#*=}" ;;
                   *) die "Unknown option: --$OPTARG" ;;
               esac ;;
            :) die "Option -$OPTARG requires an argument" ;;
            \?) die "Unknown option: -$OPTARG" ;;
        esac
    done
    shift $((OPTIND - 1))

    if [[ $# -lt 1 ]]; then
        usage
        exit "$EXIT_INVALID_ARGS"
    fi

    INPUT_FILE="$1"
}

# Validate inputs
validate() {
    [[ -f "$INPUT_FILE" ]] || die "File not found: $INPUT_FILE"
    [[ -r "$INPUT_FILE" ]] || die "File not readable: $INPUT_FILE"
    mkdir -p "$OUTPUT_DIR" || die "Cannot create output directory"
}

# Main logic
main() {
    # Defaults
    VERBOSE="${VERBOSE:-false}"
    OUTPUT_DIR="${OUTPUT_DIR:-./output}"

    parse_args "$@"
    validate

    log_info "Processing $INPUT_FILE"
    # ... main logic here ...
    log_info "Done"
}

main "$@"
```

## When to Use This Skill

- Writing new shell scripts from scratch
- Reviewing shell scripts for issues
- Refactoring legacy shell code
- Debugging script failures
- Improving script security
- Making scripts more portable
- Setting up proper error handling

Overview

This skill packages practical, modern shell scripting best practices into a concise reference you can apply immediately. It focuses on portability, strict error handling, secure coding, and readable patterns for functions, variables, and command execution. Use it to produce reliable, maintainable scripts and to review or harden existing shell code.

How this skill works

The skill inspects common shell script areas: shebang selection, strict mode, variable handling, error trapping, function design, loops, arrays, and command safety. It recommends concrete patterns (set -euo pipefail, quoting, mktemp, trap) and provides templates for logging, argument parsing, and cleanup. It highlights portability tradeoffs between POSIX and Bash and shows secure alternatives to risky constructs like eval.

When to use it

  • Creating new shell scripts that must be robust and portable
  • Reviewing or refactoring legacy shell code for maintainability
  • Adding consistent error handling and cleanup to automation scripts
  • Hardening scripts that accept user input or call external commands
  • Writing CI/CD helpers, system utilities, or admin tooling

Best practices

  • Enable strict mode: set -euo pipefail and use set -x only for debugging
  • Always quote expansions and use meaningful, readonly names for constants
  • Validate inputs early and fail fast with clear error messages and exit codes
  • Use trap to ensure cleanup of temporary files and to preserve exit codes
  • Prefer arrays and cmd arrays to avoid eval and injection; use command -v to check dependencies
  • Favor [[ ]] and (( )) in Bash, but provide POSIX fallbacks when portability is required

Example use cases

  • Template for a deploy or installer script with safe temp files, logging, and traps
  • Linting/checklist when performing a security review of scripts that handle user input
  • Refactoring a legacy sysadmin script to use functions, defaults, and explicit exit codes
  • Building CI helpers that must run across different shells and fail predictably
  • Creating small utilities that require robust argument parsing and verbose/quiet modes

FAQ

How do I choose between #!/bin/sh and #!/usr/bin/env bash?

Use /bin/sh for maximum portability and when you stick to POSIX features; use env bash when you need Bash-only features like arrays, [[ ]], or associative arrays.

What is the safest way to create temporary files?

Use mktemp or mktemp -d and register cleanup with trap to remove files/dirs on EXIT; never predict temp paths yourself.