careful

Preventing accidental data loss from destructive shell commands requires intercepting and validating terminal operations against a predefined list of dangerous patterns like recursive deletions or database drops before they execute. This skill provides that safety layer by prompting for explicit confirmation when high-risk commands are detected.

Blocking forced git pushes and recursive deletions is possible by hooking into the terminal execution flow to analyze commands in real-time. The skill detects these specific destructive patterns and requires interactive user approval to allow safe overrides or cancellation.

A Bash environment is required to intercept and validate dangerous terminal commands. The script-based safety guardrails operate within terminal-based development environments and need Bash access to hook into tool execution flows for real-time command analysis.

The best way to add safety guardrails for high-risk DevOps operations is using a command interception script that matches shell inputs against destructive patterns. This approach identifies forced git pushes and similar threats, prompting for verification to prevent accidental system damage.

Enabling destructive pattern detection when working in a shared repository involves telling the agent to be careful before beginning work. This activates the safety mode which automatically identifies dangerous operations and prompts for explicit approval prior to execution.