Scan Bash commands for destructive patterns before execution.
npx skills add https://github.com/kinetas/ai_coding_web --skill careful-kinetas
Please help me install this Agent Skill. Skill: careful Source: https://github.com/kinetas/ai_coding_web/tree/main/gstack/careful Command: npx skills add https://github.com/kinetas/ai_coding_web --skill careful-kinetas
This skill provides safety guardrails for destructive commands by warning before potentially destructive actions such as rm -rf, DROP TABLE, git reset --hard, kubectl delete, and similar operations. Users can override warnings to proceed when necessary, enabling safe debugging in production-like environments.
Attempt a command in your shell; the guard will scan it for destructive patterns and prompt you to confirm or cancel.
High-intent search queries and answers about installing and using this skill.
To prevent destructive commands like rm -rf from executing, you need a pre-tool check hook that scans input for risky patterns before execution and returns a permission decision to allow or prompt the user.
Yes, you can override safety warnings for commands like git reset --hard; the guardrail intercepts the command and prompts you to confirm or cancel, enabling safe debugging in production-like environments.
The destructive command guard works with kubectl delete, docker system prune, git push --force, DROP TABLE, TRUNCATE, and similar operations by scanning for risky patterns before execution in Bash environments.
No, build artifact directories like node_modules, .next, dist, and __pycache__ have safe exceptions that allow workflows to proceed without noise, preventing unnecessary warning prompts during development.
The best way to add guardrails for risky shell operations is hook-driven activation via PreToolUse, which enforces safety before command execution and returns a permission decision to either allow or prompt the user.
You need a pre-tool check hook for bash command safety when working in production-like environments where destructive operations such as rm -rf, git reset --hard, or kubectl delete pose risks to your data.
89jobrienConfigure Claude Code hooks for workflow automation and audit logging.
ComputerConnectionDetects destructive shell commands and prompts for confirmation via a Bash pre-execution hook.
ThalfmanConfigure git hooks to block destructive commands in Claude Code repositories.
TumeloRamaphosaDetect destructive Bash commands and request confirmation before execution.
