harness-safety-guard

To detect dangerous commands, this Skill checks command strings and code snippets against regular expression patterns to identify risky inputs like `rm -rf` or `git push --force` and blocks them before execution.

The best way to scan terminal input for destructive operations is to verify scripts and commands against known unsafe patterns, classifying threats into CRITICAL, HIGH, and MEDIUM risk levels to prevent accidental system damage.

Pattern detection classifies risk levels by matching input against regular expressions for destructive operations, assigning CRITICAL, HIGH, or MEDIUM severity to determine the appropriate safety action.

Yes, you can use this pattern detection to check cloud CLI commands like `az resource delete`, as it identifies risky patterns across code and terminal input to enforce safety protocols before execution.

A limitation of using regular expressions to identify unsafe operations is that it relies on known destructive command patterns, meaning novel or obfuscated risky code structures might not be detected without updated pattern definitions.

You need to verify a script against known destructive patterns before executing it to avoid accidental data loss, ensuring risky commands are identified and blocked prior to running.