Detect destructive commands and restrict file edits to a safe directory.
npx skills add https://github.com/a29paul/hexlens --skill guard-a29paul
Please help me install this Agent Skill. Skill: guard Source: https://github.com/a29paul/hexlens/tree/main/.claude/skills/gstack/guard Command: npx skills add https://github.com/a29paul/hexlens --skill guard-a29paul
When working on production systems or critical codebases, accidental execution of destructive commands (like rm -rf, DROP TABLE, or force-push) or unintended edits to files outside your intended work scope can cause severe outages, data loss, or irreversible damage. This skill eliminates that risk by enforcing two layers of proactive protection.
Ask the AI to activate guard mode and provide the path to the directory you want to restrict all edits to, and it will immediately enable both destructive command warnings and edit boundary enforcement.
High-intent search queries and answers about installing and using this skill.
To prevent destructive command execution, you can enable guard mode to intercept high-risk commands like rm -rf or DROP TABLE. It issues a warning before execution, allowing you to override the block only if the action is intentional.
You can lock file edits to a specific directory scope by activating directory-scoped edit restrictions. This enforces boundaries by automatically blocking any attempts to modify files outside your defined safe path.
The best way to protect production systems during live debugging is to enable integrated safety mode. This simultaneously warns you before executing destructive commands and locks edit boundaries to prevent out-of-scope file modifications.
Yes, you can override the warning when a destructive command is intentional. The safety mode alerts you before execution and provides an option to proceed, ensuring accidental data loss is prevented without blocking valid actions.
No, guard mode does not require external dependencies to enforce edit restrictions. It uses integrated pre-tool hook scripts to validate directory scope and detect destructive commands natively within your workflow.
AlejandroFiginiWarn about destructive commands and restrict edits to a designated directory.
AribSudiaEnforce a seven-phase pre-deployment verification protocol for production releases.
MybonoWrite and review bash scripts for CI/CD pipelines and system administration.
SpillwaveSolutionsConfigure hooks.yaml rules to intercept Claude Code events deterministically.
