Careful
Skill ActiveSafety guardrails for destructive commands. Warns before rm -rf, DROP TABLE, force-push, git reset --hard, kubectl delete, and similar destructive operations. User can override each warning. Use when touching prod, debugging live systems, or working in a shared environment. Use when asked to "be careful", "safety mode", "prod mode", or "careful mode". (gstack)
To prevent accidental data loss or system instability by adding a confirmation layer to dangerous shell commands, especially in production or shared environments.
Features
- Warns before destructive commands (rm -rf, DROP TABLE, force-push, etc.)
- Allows overriding warnings for proceeding with commands
- Defines safe exceptions for common patterns like deleting build artifacts
- Session-scoped activation and deactivation
Use Cases
- Use when working in production environments
- Use when debugging live systems
- Use when working in shared development environments
- Use when explicitly asked to activate 'safety mode' or 'prod mode'
Non-Goals
- Performing destructive commands directly
- Providing granular control over individual file operations
- Replacing a full-fledged shell security policy
Workflow
- User inputs a bash command.
- The 'careful' skill's hook intercepts the command before execution.
- The skill checks the command against a list of destructive patterns.
- If a destructive pattern is found, the skill warns the user and asks for confirmation.
- The user can choose to proceed or cancel the command.
Prerequisites
- Bash
- Read
Trust
- warning:Issues AttentionThere are 203 open issues and 194 closed issues in the last 90 days, suggesting a low closure rate and potentially slow maintainer response.
Code Execution
- info:LoggingThe skill includes basic local analytics logging for usage, which is helpful but not a comprehensive audit log of actions.
Installation
npx skills add garrytan/gstackRuns the Vercel skills CLI (skills.sh) via npx — needs Node.js locally and at least one installed skills-compatible agent (Claude Code, Cursor, Codex, …). Assumes the repo follows the agentskills.io format.
Quality Score
Trust Signals
Similar Extensions
Safe Mode
100Prevent destructive operations using Claude Code hooks. Three modes — cautious (warn on dangerous commands), lockdown (restrict edits to one directory), and clear (remove restrictions). Uses PreToolUse matchers for Bash, Edit, and Write.
Time Fetcher
100Instructions for fetching current Dubai time via bash command
Toprank Upgrade
100Upgrade toprank plugin to the latest version. Updates the marketplace repo, installs the new version to the plugin cache, and updates installed_plugins.json. Use when asked to "upgrade toprank", "update toprank", or "get latest version". Also handles inline upgrade prompts when a skill detects UPGRADE_AVAILABLE at startup.
Janitor Fix
100Automatically fix skill problems (safe preview first). Also use with --prune to find and remove broken symlinks, empty directories, and orphaned skills.
Trader Regime
100Detect current market regime using npx neural-trader — bull/bear/ranging/volatile classification with recommended strategy
Setup
100Use first for install/update routing — sends setup, doctor, or MCP requests to the correct OMC setup flow