skills.rest
glfruit
glfruitCOMMUNITY
Explore all skills by this author →
WebsiteGitHub

safe-mode

Enable per-session safety modes that constrain destructive operations via activation keywords.

Updated Feb 6, 2026
Search Tags:#read-only#confirmation#ai-safety#dry-run#session-control#safety-mode
One-click install
npx skills add https://github.com/glfruit/gorin-skills --skill safe-mode
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: safe-mode
Source: https://github.com/glfruit/gorin-skills/tree/main/openclaw/safe-mode
Command: npx skills add https://github.com/glfruit/gorin-skills --skill safe-mode

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Session-scoped safety modes provide temporary, per-session constraints to prevent destructive or sensitive operations in AI interactions, improving control and reducing accidents.

Core Features & Use Cases

  • Activation via keywords (/careful, /freeze, /dry-run) to apply constraints for the current session.
  • Read-only and confirmation modes that block or gate risky actions while allowing safe operations.
  • Use Case: When testing a complex prompt, enable safety mode to review commands and outputs before execution.

Quick Start

Activate safety mode by sending /careful, /freeze, or /dry-run in your message to apply constraints for the current session.

Frequently Asked Questions about safe-mode

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I enable a read-only mode for an AI session to prevent destructive actions?▼

You can activate read-only mode by sending specific keywords like /careful, /freeze, or /dry-run in your message to apply temporary constraints for the current session, blocking risky actions while allowing safe operations.

What is session-scoped safety mode in AI interactions?▼

Session-scoped safety mode provides temporary, per-session constraints that prevent destructive or sensitive operations in AI interactions, improving control and reducing accidents during complex tasks.

How do I set up a dry-run to review commands before execution?▼

Trigger a dry-run by sending the /dry-run keyword in your message; this activates safety constraints for the current session, allowing you to review commands and outputs before any execution occurs.

Can I use confirmation modes to gate risky actions while testing complex prompts?▼

Yes, confirmation modes block or gate risky actions while allowing safe operations, making them ideal when testing complex prompts to review commands and outputs before execution.

When should I not use safety mode for write-like actions?▼

You should not use safety mode for write-like actions when you need immediate, unchecked execution without confirmation gates, as the session control blocks or delays risky operations until constraints expire or are deactivated.

Related Skills

Explore More Software Engineering Skills→
mindfold-aimindfold-ai
13.8k

trellis-bootstrap

Initialize the Trellis AI workflow system in a new project.

Official
Advanced
haojing8312haojing8312
154

workclaw-cleanup-execution

Execute reviewed repository hygiene cleanup batches and route verification through workclaw-change-verification.

Community
Intermediate
brolagbrolag
11

prompt-engineering

Analyzes and improves AI prompts using the CRISP-E framework.

Community
Intermediate
CruxExpertsCruxExperts
3

ls-agentic-umbrella-queue

Enforce scope-aware umbrella workflows with explicit user confirmation before destructive actions.

Community
Advanced