gh-dotclaude-marketplace-plugins-cli-mastery/commands/shell.md

model, allowed-tools, argument-hint, description

model	allowed-tools	argument-hint	description
claude-sonnet-4-0	Task, Bash, Read, Write	<script-purpose> [robustness]	Shell script creation with error handling and best practices

Shell Command

Shell script creation with error handling and best practices

SECURITY WARNING

CRITICAL: This command creates shell scripts with Bash execution capabilities.

Scripts you create will have the power to:

• Execute system commands
• Modify/delete files
• Access network resources
• Change permissions and ownership
• Potentially escalate privileges

BEFORE requesting a script, consider:

• What's the blast radius if this script is exploited?
• Does this script handle any untrusted input?
• Will this script access sensitive data or credentials?
• What's the minimum privilege level needed?

Security Requirements Checklist

EVERY shell script must include:

• Strict error handling: set -euo pipefail
• Input validation: Validate ALL external inputs with regex
• Quoted variables: ALWAYS quote variables to prevent injection
• No hardcoded secrets: Use env vars or secret management
• Secure temp files: Use mktemp, never predictable names
• Least privilege: Run with minimum necessary permissions
• Cleanup handlers: Use trap to clean up on exit/error
• Safe file permissions: chmod 700 for scripts, 600 for configs
• Command validation: Validate commands before execution
• Audit logging: Log security-relevant operations

Dangerous Operations to Avoid

STOP and think before using:

• rm -rf with variables
• chmod 777 or similar overly permissive modes
• sudo without specific command limits
• eval with external input
• Unquoted variables in commands
• Shell injection via unsanitized input

Arguments

$1 (Required): script-purpose

$2 (Optional): robustness (production requires strict security)

Examples

/shell "Backup database with rotation" production
/shell "Deploy application with health checks"

Invoke the shell-scripter agent with: $ARGUMENTS