Can it prevent system-level damage?

Yes, Class X (Catastrophic) actions such as root/system destruction or disk-boot operations are automatically refused regardless of confirmation.

What is workspace-only scoping?

It ensures the skill only operates within the git top-level or current directory, rejecting any mutation targets or symlink escapes outside the defined workspace.

How does this skill handle destructive commands?

It classifies them as Class 2 risk and requires an explicit 'APPROVE-DESTRUCTIVE' payload confirmation and a safer alternative preview before execution.

Does it protect my API keys and credentials?

Absolutely; the C3 Secret Safety policy ensures that secret values, keys, and raw authentication headers are never emitted or logged.

ChatGPT CLI Safety Guardrails

Name: ChatGPT CLI Safety Guardrails
Author: AndrewRober

byAndrewRober

0•

安全与测试

Enforces security policies and safety guardrails for ChatGPT CLI operations to prevent destructive actions and credential leaks.

This skill acts as a comprehensive safety layer for users interacting with ChatGPT via the command line, implementing a structured risk-classification system from read-only to catastrophic. It prevents high-risk system commands, mandates explicit user approval for destructive file mutations or cloud-based operations, and actively masks sensitive secrets to prevent leakage. By enforcing workspace-only scoping and requiring safer alternatives like 'dry-runs' for certain commands, it minimizes the risk of accidental data loss or unauthorized external execution while maintaining a transparent audit trail of approved actions.

主要功能

01Prevention of catastrophic system, disk, and security-control operations

02Hierarchical risk classification from Class 0 (read-only) to Class X (refusal)

03Strict workspace scoping to prevent file mutations outside of the project root

04Mandatory approval prompts for cloud-based mutations and destructive local changes

05Automated secret safety to prevent the emission of keys and auth headers

060 GitHub stars

使用场景

01Securing cloud infrastructure deployments by requiring explicit payload verification

02Preventing accidental recursive file deletion or moves during automated refactoring

03Sanitizing CLI output to ensure sensitive API credentials are never exposed

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add andrewrober/codexcli-guardrails guard-users-chatgpt

For use in Claude.ai and ChatGPT

Download Skill