DicklesworthstoneDicklesworthstoneCommunity·1 Agent Skills Included

slb

Peer review and approval for destructive terminal commands

Requires a second reviewer to approve risky commands like rm -rf, git push --force, or DROP TABLE before they run. Classifies commands into risk tiers and blocks dangerous operations until explicit approval is granted. Prevents irreversible mistakes from hallucinating or tunnel-visioned coding sessions, with full audit logs and a terminal dashboard for human reviewers.
npx skills add Dicklesworthstone/slb --all -g -y
Available:

Sets strict safety rules for the agent working in this codebase, forbidding file deletion and destructive git commands without explicit written user permission, and defining build, test, and formatting requirements.

All Skills in This Repository (1)

Pure Emerald Level Indicators

Frequently Asked Questions

FAQPage Schema
How to install slb?

Run `npx skills add Dicklesworthstone/slb --all -g -y` in your terminal to install this skill globally.

What does slb protect against?

It blocks destructive commands like rm -rf, git reset --hard, kubectl delete, and DROP TABLE until a second reviewer approves them, preventing irreversible damage from agent mistakes.

How does the two-person rule work in slb?

An agent submits a risky command with a reason, a second agent or human reviews and approves it, and only then does the command execute in the original shell environment.

Does slb work with Claude Code and Cursor?

Yes. It generates Claude Code hooks and Cursor rules so dangerous commands are automatically routed through the approval workflow.

Can humans review pending commands in slb?

Yes. It includes an interactive terminal dashboard where human reviewers can inspect, approve, or reject pending requests, plus an emergency override with full logging.

Related Repositories in Software Engineering

View All in Software Engineering