Full-Stack Quality Skill

A reusable AI coding skill for reviewing, designing, and implementing maintainable full-stack software.

It covers database design, backend architecture, frontend architecture, mobile, HTTP APIs, GraphQL, security, privacy, accessibility, i18n, analytics, background jobs, AI/LLM safety, infrastructure, documentation, multi-tenancy, payments/billing, notifications, data import/export, API compatibility, developer experience, utilities, testing, performance, observability, and delivery/configuration.

What This Skill Is For

• Read-only code reviews and audits.
• Full-stack implementation guidance.
• Architecture and ADR support.
• Database normalization and migration review.
• API design review for REST/HTTP and GraphQL.
• Frontend framework guidance for React, Vue, Angular, Svelte, and server-rendered templates.
• Security, testing, performance, observability, and CI/CD hygiene.
• Optional project context mode: audit repo, decide target architecture, generate project docs, verify drift, track long-term memory, health score, and CI quality gates.

Core Rule

Audit/review mode is read-only. The agent may recommend changes, but must not edit source code, schemas, configs, docs, generated files, or lockfiles unless the user explicitly asks for implementation or document generation.

Coverage

Core areas:

• database, backend, frontend, mobile;
• HTTP API, GraphQL, API compatibility;
• security, privacy, accessibility, i18n;
• testing, performance, observability, delivery/configuration;
• utilities, documentation, developer experience;
• analytics, background jobs, notifications;
• multi-tenancy, payments/billing, data import/export;
• AI/LLM safety and infrastructure/cloud/IaC.

The skill is intentionally stack-aware but language-agnostic. It should preserve the project's existing architecture and conventions unless the user asks to change them.

Folder Structure

full-stack-quality/
├── SKILL.md
├── README.md
├── agents/
│   ├── openai.yaml
│   ├── codex-AGENTS.md
│   ├── claude-CLAUDE.md
│   ├── cursor-full-stack-quality.mdc
│   ├── antigravity-full-stack-quality.md
│   ├── gemini-AGENTS.md
│   └── windsurf-full-stack-quality.md
├── auditors/
├── examples/              # stack/language examples, intentionally selective
├── references/
└── templates/

Examples

The examples/ folder is intentionally selective. It gives the agent enough patterns to translate into most stacks without becoming a 50-language cookbook.

Backend examples:

• Node.js / TypeScript
• Python
• Django
• PHP / Laravel
• Java / Spring
• C# / ASP.NET Core
• Go
• Ruby on Rails

Frontend and full-stack examples:

• React
• Next.js
• Vue
• Angular
• SvelteKit

Mobile examples:

• Flutter
• React Native
• Kotlin / Android
• Swift / iOS

API/data examples:

• SQL
• HTTP API
• GraphQL

If the exact language is missing, read examples/language-selection.md and translate the closest pattern into the repo's style.

Installation

Codex / OpenAI-Style Agents

Codex-style agents can use the skill through AGENTS.md:

git submodule add https://github.com/lablnet/full-stack-quality-skill.git tools/full-stack-quality

Add this to the project's AGENTS.md:

For full-stack quality reviews, read tools/full-stack-quality/SKILL.md
and follow it. Supporting files are in the same folder.

Use it with prompts like:

Use full-stack-quality to review this project. Audit only, read-only.

Use full-stack-quality to review the backend, database schema, and HTTP API. Produce review.md and findings.json.

Claude Code

For all projects, install the skill in your personal Claude skills directory:

git clone https://github.com/lablnet/full-stack-quality-skill.git ~/.claude/skills/full-stack-quality

For one project, add it inside the repository:

git submodule add https://github.com/lablnet/full-stack-quality-skill.git .claude/skills/full-stack-quality

If the project needs persistent Claude instructions, copy or merge:

cp .claude/skills/full-stack-quality/agents/claude-CLAUDE.md CLAUDE.md

Use it with prompts like:

Use the full-stack-quality rules to audit this repo. Do not edit files.

Cursor

Install the skill inside Cursor's project tools folder:

git submodule add https://github.com/lablnet/full-stack-quality-skill.git .cursor/tools/full-stack-quality

Create .cursor/rules/full-stack-quality.mdc:

---
description: Full-stack quality review for database, backend, frontend, APIs, GraphQL, security, testing, performance, observability, delivery, and utilities.
---

Read .cursor/tools/full-stack-quality/SKILL.md and follow it.

Use it with:

Apply the full-stack-quality rule and review the frontend architecture.

To update the Cursor submodule later:

git submodule update --remote .cursor/tools/full-stack-quality
git add .cursor/tools/full-stack-quality
git commit -m "Update full-stack-quality skill"

Antigravity

Antigravity supports workspace rules under .agents/rules/ and workspace skills under .agents/skills/.

For a workspace skill:

git submodule add https://github.com/lablnet/full-stack-quality-skill.git .agents/skills/full-stack-quality

Optional workspace rule in .agents/rules/full-stack-quality.md:

Read .agents/skills/full-stack-quality/SKILL.md and follow it for full-stack quality review and implementation tasks.

Use it with:

Use the full-stack-quality skill to review this project read-only.

Gemini Tools That Support AGENTS.md

Install the skill as a project tool:

git submodule add https://github.com/lablnet/full-stack-quality-skill.git tools/full-stack-quality

Add this to AGENTS.md:

For full-stack quality reviews, read tools/full-stack-quality/SKILL.md
and follow it. Supporting files are in the same folder.

Use it with:

Use the AGENTS.md full-stack quality rules to audit this codebase.

Windsurf

Install the skill as a project tool:

git submodule add https://github.com/lablnet/full-stack-quality-skill.git tools/full-stack-quality

Create .windsurf/rules/full-stack-quality.md:

Read tools/full-stack-quality/SKILL.md and follow it for full-stack quality review and implementation tasks.

If your setup uses the legacy root .windsurfrules file, add the same instruction there.

Use it with:

Review this repo using the full-stack-quality rules. Audit only.

Output Formats

• Markdown for human review reports, explanations, ADRs, and implementation plans.
• JSON for structured findings and machine-readable checklists.
• YAML only when the existing project/tool requires YAML or the user asks for it.

Preferred output for single-area audits:

findings.json
review.md

Required output for broad, multi-area, or parallel audits unless the user explicitly asks for one combined file:

review.md
findings/
├── database.findings.json
├── backend.findings.json
├── frontend.findings.json
├── http-api.findings.json
├── graphql.findings.json
├── security.findings.json
├── utilities.findings.json
├── testing.findings.json
├── performance.findings.json
├── observability.findings.json
├── delivery.findings.json
├── mobile.findings.json
├── accessibility.findings.json
├── privacy.findings.json
├── multi-tenancy.findings.json
├── payments.findings.json
├── notifications.findings.json
├── data-import-export.findings.json
├── api-compatibility.findings.json
└── developer-experience.findings.json

Use one findings.json only for single-area reviews or when the user explicitly asks for one combined file. Use per-area JSON files when more than one auditor runs.

Parallel Audits

For whole-project audits, the skill asks agents with subagent support to launch one read-only auditor per relevant area in parallel. It should skip irrelevant areas, such as GraphQL when the project has no GraphQL schema, resolver, operation, or dependency.

For narrow reviews, it should use only the matching auditor. A frontend-only review should not launch every auditor.

Project Context Mode

For deeper repository work, the skill can also:

• audit the repo;
• decide a target architecture;
• generate docs that guide future agents;
• keep long-term project memory;
• compute health score and trend;
• verify drift;
• support CI quality gates.

Recommended files:

docs/
├── agents.md
├── architecture.md
├── decisions.md
├── security.md
├── testing.md
├── migration-backlog.md
└── audit/
    ├── inventory.json
    ├── drift-report.md
    └── findings/
        ├── backend.findings.json
        └── security.findings.json

Health score is a trend signal, not an absolute truth. Keep the formula stable so future runs can be compared.

Agent Tool Adapters

The agents/ folder contains copyable starter files for popular AI coding tools:

• Codex/OpenAI-style agents: copy agents/codex-AGENTS.md to AGENTS.md.
• Claude Code: copy agents/claude-CLAUDE.md to CLAUDE.md.
• Cursor: copy agents/cursor-full-stack-quality.mdc to .cursor/rules/full-stack-quality.mdc.
• Antigravity: copy agents/antigravity-full-stack-quality.md to .agents/rules/full-stack-quality.md.
• Gemini tools that support agent files: copy agents/gemini-AGENTS.md to AGENTS.md.
• Windsurf: copy agents/windsurf-full-stack-quality.md to .windsurf/rules/full-stack-quality.md or adapt it to the project's Windsurf rules format.