home / skills / beshkenadze / claude-skills-marketplace / safe-rm

safe-rm skill

safe

This skill helps you safely remove files and directories by wrapping rm -rf with protective checks and dry-run previews.

npx playbooks add skill beshkenadze/claude-skills-marketplace --skill safe-rm

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

Files (2)

SKILL.md

1.3 KB

---
name: safe-rm
description: Use safe-rm script instead of rm -rf. Prevents accidental deletion of protected system, config, and project paths.
version: 1.0.0
---

# Safe RM

## Overview

Wrapper around `rm -rf` with safety checks. Use this instead of direct `rm -rf` commands.

## When to Use

- Deleting directories or files recursively
- Any `rm -rf` or `rm -r` operation
- Cleaning up temporary files/folders

## Usage

```bash
# Dry-run (default) - shows what would be deleted
safe-rm /path/to/delete

# Actually delete (if path is not protected)
safe-rm --force /path/to/delete
safe-rm -f /path/to/delete
```

## Protected Paths

The script blocks deletion of:

**System:** `/`, `/bin`, `/boot`, `/dev`, `/etc`, `/lib`, `/opt`, `/proc`, `/root`, `/sbin`, `/sys`, `/usr`, `/var`, `/home`, `/Users`

**Home config:** `~`, `~/.ssh`, `~/.gnupg`, `~/.config`, `~/.local`, `~/.claude`, `~/.zshrc`, `~/.bashrc`

**Project:** `.git`, git repository root

## Exit Codes

- `0` — success (or dry-run completed)
- `1` — path is protected
- `2` — path does not exist

## Instructions for Claude

1. **NEVER use `rm -rf` directly** — always use `safe-rm` script
2. Run without `--force` first to preview what will be deleted
3. If path is protected, inform user and do not proceed
4. Only use `--force` after confirming dry-run output is correct

Overview

This skill provides a safe-rm wrapper around rm -rf to prevent accidental deletion of critical system, configuration, and project paths. It enforces a dry-run by default and blocks protected locations, requiring an explicit --force to perform removals. The goal is to make destructive file operations deliberate and auditable.

How this skill works

safe-rm inspects target paths before invoking removal: it checks existence, matches against a list of protected system and home config locations, and detects common project markers (like .git or repo root). By default it performs a dry-run showing what would be removed; only when --force is provided and no protection matches will it execute rm -rf. It exits with meaningful codes: 0 for success/dry-run, 1 for protected path, and 2 for missing path.

When to use it

Any recursive deletion (rm -r / rm -rf)
Cleaning temporary files or build artifacts
When removing directories that may contain config or hidden files
Before running batch delete scripts to verify targets
When working in repositories where .git or repo root should be preserved

Best practices

Always run safe-rm without --force first to preview deletions
Review dry-run output carefully before confirming destructive actions
Never bypass protections for system or home config paths
Use safe-rm in automation and CI to reduce catastrophic mistakes
Keep the protected-paths list updated to reflect your environment

Example use cases

Safely remove a large build directory after confirming contents: safe-rm /path/to/build
Clean a temp workspace in CI with a dry-run step, then use --force when approved
Avoid accidentally deleting your home dotfiles by blocking ~/.ssh and shell rc files
Protect repository integrity by preventing deletion at repo root or .git
Use in developer scripts to replace direct rm -rf calls for safer operations

FAQ

What happens if I run safe-rm on a protected path?

The script aborts, returns exit code 1, and prints a clear message identifying the protection; no deletion occurs.

How do I actually remove files after a dry-run?

Re-run the same safe-rm command with --force or -f after confirming the dry-run output; safe-rm will re-check protections before proceeding.