Intercept destructive commands with Careful, Freeze, and Guard modes.
npx skills add https://github.com/arbisoft/ai-skillforge --skill safety-guard-arbisoft
Please help me install this Agent Skill. Skill: safety-guard Source: https://github.com/arbisoft/ai-skillforge/tree/main/Claude/skills/safety-guard Command: npx skills add https://github.com/arbisoft/ai-skillforge --skill safety-guard-arbisoft
Use this skill to prevent destructive operations when working on production systems or running autonomous agents.
Enable safety-guard in the active session and attempt a potentially destructive command to observe the protective prompts.
High-intent search queries and answers about installing and using this skill.
To prevent destructive commands in production environments, a safety guard skill intercepts risky operations before execution. It applies PreToolUse hooks to catch potentially harmful commands during autonomous agent runs and admin operations, requiring confirmation before proceeding.
Yes, you can lock file edits to a specific directory tree using Freeze Mode. This mode prevents unintended changes outside a target path by restricting write access, while still allowing read access to other areas.
The best way to add guardrails for autonomous agents is by enabling Guard Mode, which combines multiple protections for maximum safety. It allows read access but restricts writes to designated areas, intercepting destructive commands before they execute.
To log blocked actions when intercepting destructive commands, the skill uses PreToolUse hooks. These hooks automatically intercept potentially harmful commands and log the blocked actions, providing an audit trail of prevented operations.
Yes, safety-guard works with codex sessions and aligns with ECC risk controls. It includes integration guidance for enabling safety mechanisms during these sessions, ensuring destructive operations are intercepted and prevented.
blacklanternsecurityAutomate AD CS persistence via Golden Certificate forging and certificate theft.
emiljuchnikowskiParse Claude Code audit logs and aggregate events by type.
ricableBuilds graph-based autonomous AI agent workflows with webhook and schedule triggers.
JamkrisIntercept destructive commands and block unauthorized edits via PreToolUse hooks.
