skills.rest
adamtpang
adamtpangCOMMUNITY
Explore all skills by this author →
Website𝕏@adamtpangGitHub

guard

Blocks destructive shell commands and restricts file edits to a chosen directory.

Updated Jun 16, 2026
Search Tags:#production#hooks#safety#freeze#guard#edit boundary#destructive commands
One-click install
npx skills add https://github.com/adamtpang/summon.company --skill guard-adamtpang
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: guard
Source: https://github.com/adamtpang/summon.company/tree/main/.claude/skills/guard
Command: npx skills add https://github.com/adamtpang/summon.company --skill guard-adamtpang

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? When working on production systems or live environments, a single destructive command (rm -rf, DROP TABLE, force-push) or an edit to the wrong file can cause serious damage. This Skill activates a combined safety mode that warns before dangerous commands and blocks edits outside a directory you specify. ## Core Features & Use Cases - Destructive Command Warnings: PreToolUse hooks intercept Bash commands matching dangerous patterns (rm -rf, DROP TABLE, force-push) and warn before execution, with the option to override. - Directory-Scoped Edit Boundary: Edit and Write operations are restricted to a user-specified directory; changes outside the freeze boundary are blocked. - Use Case: While debugging a live production service, you tell the agent to lock edits to /srv/app/config. Any attempt to modify files elsewhere is blocked, and risky shell commands trigger a confirmation warning first. ## Quick Start Ask the agent to activate guard mode and restrict edits to your target directory, for example: enable full safety mode and only allow edits under /srv/app/config.

Frequently Asked Questions about guard

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

FAQPage Schema
How do I block file edits outside a specific directory?▼

Activate guard mode and provide the target directory path when prompted. The skill resolves it to an absolute path, saves it to a freeze state file, and PreToolUse hooks then block Edit and Write operations outside that boundary.

How do I get warnings before running destructive shell commands?▼

Guard mode registers a PreToolUse hook on Bash that checks commands against destructive patterns such as rm -rf, DROP TABLE, and force-push. Matching commands trigger a warning before execution, and you can choose to override.

Does guard mode work without the careful and freeze skills installed?▼

No. Guard depends on hook scripts from the sibling careful and freeze skill directories under gstack. Both must be installed, which the gstack setup script handles automatically.

How do I turn off the edit boundary after enabling guard mode?▼

Run the /unfreeze command to remove the directory edit restriction. To deactivate all protections entirely, end the current session, since the hooks are session-scoped.

Can I override a destructive command warning in guard mode?▼

Yes. Destructive command checks produce warnings rather than hard blocks, so you can confirm and proceed. Only the directory edit boundary is strictly enforced as a block.

Related Skills

Explore More Software Engineering Skills→
sanjanbsanjanb
1

guard

Warns on destructive bash commands and restricts file edits to a chosen directory.

Community
Intermediate
VertaKhanVertaKhan

unfreeze

Clears the freeze boundary state file to allow edits in all directories again.

Community
Basic
dnsnamdangdnsnamdang

Entity History

Log entity record modifications with user information and timestamps in PHP.

Community
Advanced
calebeynoncalebeynon

docker-sandbox

Orchestrate isolated Claude Code sessions in Docker sandboxes with git branch restrictions.

Community
Advanced