Builderoom
Get resource

Quality-check gates for AI coding agents covering code, tests, and docs—catch common generation flaws before commit.

MITClaude CodeCursorText InputCode InputText Output
安装 / 调用方式bash
```bash
# 浏览可用 skills
npx skills add amElnagdy/guard-skills --list

# 安装全部
npx skills add amElnagdy/guard-skills

# 安装单个 guard
npx skills add amElnagdy/guard-skills --skill clean-code-guard

# 为特定代理安装
npx skills add amElnagdy/guard-skills --skill test-guard --agent claude-code

# 全局安装
npx skills add amElnagdy/guard-skills --global
```
调用示例json
```
Use $clean-code-guard on the diff you just produced.
Use $test-guard on the tests you just wrote.
Use $docs-guard on this README update before we ship it.
Use $wp-guard on this WordPress plugin change.
Use $woo-guard on this WooCommerce checkout change.
```

Product facts

Type
AI Skill & Agent
Category
AI & Machine Learning › Prompt Engineering & Skill Management
Pricing
Open source
Platforms
Claude Code · Cursor
License
MIT
Launched
Sep 26, 2026
Listed
Sep 26, 2026
Website
github.com

Overview

GitHubGitHub

guard-skills is a set of quality-gate skills for AI coding agents that double-check AI-generated code, tests, and documentation before they are committed or merged. ## What problem it solves AI coding agents have systematic failure modes when generating code: over-abstraction, swallowed errors, hallucinated APIs, test bloat, and docs drifting out of sync with code. guard-skills acts as a reactive review step, inspecting the diff after the agent finishes its work to catch these issues. ## Key capabilities The repo contains five standalone guard skills: - **clean-code-guard**: For production code in any language. Checks for common LLM code smells, over-abstraction, overly broad error catching, naming issues, and SOLID/DRY/KISS/YAGNI violations. - **test-guard**: For test code. Supports frameworks like pytest, PHPUnit/Pest, Jest/Vitest, and Go tests. Checks for mock abuse, duplicate tests, assertions on implementation details, and more. - **docs-guard**: For READMEs, API docs, docstrings, changelogs, etc. Treats documentation as a set of claims and validates them against the codebase, checking for hallucinated symbols, broken examples, and doc drift. - **wp-guard**: For WordPress plugins, themes, REST endpoints, AJAX handlers, etc. Checks escaping, sanitization, nonces, capabilities, prepared queries, i18n, and caching issues. - **woo-guard**: For WooCommerce extensions, checkout, order, and product logic. Builds on wp-guard to check HPOS compatibility, direct order meta access, amount handling, and more. ## Who it's for Developers using AI coding agents such as Claude Code, Codex, Cursor, or OpenCode. Best practice: let the agent finish its work first, then invoke the appropriate guard to review the diff before presenting, committing, or merging. ## Technical highlights Each skill consists of a `SKILL.md` entry file, lightweight agent metadata, and progressively disclosed reference docs. No executable scripts, network calls, MCP server dependencies, or credentials—everything is fully inspectable. Install and manage via Skills CLI.

Key features

  • clean-code-guard: Checks production code in any language for LLM code smells, over-abstraction, broad error catching, and SOLID violations.
  • test-guard: Checks test code for nine categories of issues including mock abuse, duplicate tests, and assertions on implementation details.
  • docs-guard: Treats docs as claims validated against the codebase, catching hallucinated symbols, broken examples, and doc drift.
  • wp-guard: Checks WordPress code for escaping, sanitization, nonces, capabilities, prepared queries, and i18n issues.
  • woo-guard: Checks WooCommerce code for HPOS compatibility, direct meta access, and amount handling issues.
  • Multi-agent support: Works with AI coding agents like Claude Code, Codex, Cursor, and OpenCode.
  • Skills CLI install: Install all or individual guards via `npx skills add`, supporting global and project-level installs.
  • Zero execution risk: Markdown and YAML metadata only—no executable scripts, network calls, or MCP server dependencies.
  • Progressive disclosure: `SKILL.md` stays lean to reduce load cost; deep reference docs load only when the task requires them.

Releases

Notify me

Initial · guard-skills

Q&A

How do I install guard-skills?AI-generated
Install via Skills CLI. All skills: `npx skills add amElnagdy/guard-skills`. Single skill: `npx skills add amElnagdy/guard-skills --skill clean-code-guard`. You can also target an agent with `--agent codex` or install globally with `--global`.
When should I use these guards?AI-generated
Best practice is to let the agent finish its work first, then invoke the appropriate guard to review the generated diff before presenting, committing, or merging. You can also explicitly request guard constraints while writing, but they work best as a reactive review step.
How do I update installed skills?AI-generated
Skills are installed as copies, so new versions don't sync automatically. Run `npx skills update` to update all installed skills, or `npx skills update clean-code-guard` for a specific one. Re-running the add command also pulls the latest version.
Which AI coding agents are supported?AI-generated
Claude Code, Codex, Cursor, OpenCode, and other agents supported via Skills CLI.
Do these skills contain executable code?AI-generated
No. Skill content is Markdown plus lightweight `agents/openai.yaml` metadata—no executable scripts, network calls, MCP server dependencies, or credentials.
What's the difference between wp-guard and woo-guard?AI-generated
wp-guard covers WordPress-level security checks (escaping, nonces, capabilities, etc.). woo-guard builds on wp-guard to handle WooCommerce-specific concerns like HPOS compatibility, CRUD over direct meta operations, checkout validation, and amount handling.
More
Get resource
Bookmark

This product has no owner yet

It was listed by the platform or its author left. If it is your work, prove your relationship to become the owner.

Claim this product

Alternatives

The product content on this page was translated by AI from the author's original.