guard

Orchestrate pre-commit safety reviews of staged git changes via a haiku-based agent.

Updated Apr 9, 2026
One-click install
npx skills add https://github.com/HurleySk/claude-cicd-safety-skill --skill guard-hurleysk
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: guard
Source: https://github.com/HurleySk/claude-cicd-safety-skill/tree/main/skills/guard
Command: npx skills add https://github.com/HurleySk/claude-cicd-safety-skill --skill guard-hurleysk

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

The Guard provides a pre-commit safety review workflow that reminds and orchestrates a deliberate inspection of risky changes before commits happen, reducing production risk.

Core Features & Use Cases

  • PostToolUse hook that detects risky commits and prints a reminder to invoke a Guard review.
  • Manual Guard review workflow dispatched via the haiku agent for fast, targeted analysis of the full changeset.
  • Configurable safety rules from .claude/safety-rules.json to tailor production indicators and URIs for project-specific checks.
  • Non-blocking safety guardrails: the system prompts the user to take action rather than blocking the commit automatically.

Quick Start

Commit changes in a project to trigger a Guard reminder and then run the Guard review with the /cicd-safety:guard review command.

Frequently Asked Questions about guard

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

FAQPage Schema
How do I run a pre-commit safety review for staged git changes?

You can run a pre-commit safety review by triggering the PostToolUse hook and executing the Guard review command to assess staged git changes for production-risk indicators before finalizing your commit.

Does the guard skill automatically block risky commits in my CI/CD workflow?

No, the guard skill provides non-blocking safety guardrails in your CI/CD workflow, prompting the user to take manual action and review risky changes rather than automatically blocking the commit.

How do I configure project-specific production risk indicators for git commits?

You can configure project-specific production risk indicators by defining custom safety rules in the .claude/safety-rules.json file to tailor the Guard review assessment.

What is the best way to inspect risky code changes before they reach production?

The best way is to orchestrate a deliberate pre-commit safety review using the haiku agent, which performs fast, targeted analysis of the full changeset to reduce production risk.

Do I need a post-commit hook to use the guard safety review?

Yes, the guard safety review requires a post-commit hook to detect risky commits and print a reminder, along with a haiku-based agent prompt system to execute the review.

Why use a non-blocking safety review instead of automated hard blocks for git commits?

A non-blocking safety review guides teams to apply review-only safeguards, ensuring developers remain aware of production-risk indicators without disrupting automated CI/CD pipelines with hard blocks.