What commands are allowed to run automatically?

Safe, read-only commands such as git status, git diff, ls, cat, and find are permitted to run without manual confirmation.

Can this prevent accidental force pushes to GitHub?

Absolutely. Any attempt to use force flags in Git is intercepted, requiring you to manually verify the action first.

Which commands require explicit user approval?

Destructive or high-impact operations like rm -rf, git reset --hard, git push --force, sudo commands, and shell piping require confirmation.

Does this skill protect sensitive configuration files?

Yes, the skill triggers a confirmation requirement whenever Claude attempts to create or modify .env files or other secret-containing documents.

How does it handle package managers like pnpm?

For commands like pnpm install or pnpm lint:fix, the skill requires Claude to provide an explanation of why the command is necessary before execution.

Safe Command Policy

Name: Safe Command Policy
Author: Hakuya5247

byHakuya5247

0•

보안 및 테스팅

Enforces a safety layer for terminal command execution by requiring explicit user confirmation for destructive or high-impact operations.

Safe Command Policy is a specialized skill for Claude Code designed to minimize risk during CLI-driven development. It establishes a clear hierarchy of command permissions, allowing benign operations like git status and file reading to proceed automatically while acting as a gatekeeper for potentially dangerous actions. By intercepting high-risk commands such as hard resets, force pushes, and administrative operations, this skill ensures that Claude never executes a destructive task without human oversight, providing a robust safety net for your local environment and remote repositories.

주요 기능

01Mandatory confirmation for dangerous commands including rm -rf and sudo

02Protection against accidental git reset --hard and git push --force

03Automatic execution of non-destructive read operations like ls and git status

04Security guardrails for .env files and sensitive credential management

050 GitHub stars

06Requirement for contextual explanation before running package manager updates

사용 사례

01Preventing accidental data loss or repository corruption during complex Git workflows

02Securing local environments from unauthorized sudo or curl-to-shell operations

03Standardizing a human-in-the-loop verification process for production-impacting scripts

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add hakuya5247/yomicord safe-command-policy

For use in Claude.ai and ChatGPT

주요 기능

01Mandatory confirmation for dangerous commands including rm -rf and sudo

02Protection against accidental git reset --hard and git push --force

03Automatic execution of non-destructive read operations like ls and git status

04Security guardrails for .env files and sensitive credential management

050 GitHub stars

06Requirement for contextual explanation before running package manager updates

사용 사례

01Preventing accidental data loss or repository corruption during complex Git workflows

02Securing local environments from unauthorized sudo or curl-to-shell operations

03Standardizing a human-in-the-loop verification process for production-impacting scripts