required-reading

Claude knows every engineering principle. This plugin makes it enforce them.

A Claude Code plugin that transforms Claude from a permissive code generator into a full-lifecycle engineering standards enforcer — covering 10 domains, with checklists that ensure nothing gets skipped, and a multi-agent team mode that spins up domain specialists for thorough reviews.

What It Does

Claude defaults to matching whatever style the codebase uses, staying quiet about violations, and being "helpful" by writing code that works but accumulates technical debt.

required-reading changes that. Once installed, Claude will:

• Enforce standards across 10 domains — not just code quality, but architecture, testing, security, DevOps, data, delivery, product, UX, and leadership
• Run pre-delivery checklists — triggered automatically by work type (writing code, reviewing, designing, deploying, etc.)
• Flag anti-patterns on sight — God classes, anemic domain models, N+1 queries, hardcoded secrets, flaky tests, and dozens more
• Launch specialist subagents — Team Mode spins up parallel domain experts for comprehensive multi-domain reviews
• Be pragmatic — scope-appropriate enforcement, prototype-aware, performance-conscious, legacy-friendly

What It Covers

| Domain | Key Standards | |--------|--------------| | Software Engineering | Naming, function design, CQS, error handling, refactoring, legacy code strategy | | Architecture & Design | SOLID, deep modules, dependency rule, bounded contexts, DDD, design patterns | | QA & Testing | TDD, test pyramid, test design, test doubles, boundary analysis, no flaky tests | | Security | Input validation, output encoding, auth/authz, secrets management, STRIDE threat modeling | | DevOps & Reliability | CI/CD, IaC, deployment strategies, observability, SLOs, stability patterns | | Data Engineering | Data modeling, indexing, query optimization, pipelines, schema migrations, data quality | | Delivery & Process | Small batches, WIP limits, estimation, Definition of Done, tech debt visibility | | Product Management | Outcomes over outputs, validated learning, continuous discovery, story mapping | | UX Engineering | Usability, WCAG 2.1 AA, Core Web Vitals, atomic design, design tokens | | Technical Leadership | ADRs, psychological safety, intent-based leadership, capacity allocation |

Installation

Quick Install

npx required-reading

Installs the core skill + all 10 domain specialists to ~/.claude/skills/. Works on macOS, Linux, and Windows.

Claude Code Plugin

/plugin marketplace add Drmsay/Required-Reading
/plugin install required-reading@required-reading

Per-Project

Copy the contents of claude-md-snippet.md into your project's CLAUDE.md for a condensed version scoped to a single project.

Checklists

Checklists trigger automatically based on the type of work:

| Work Type | Triggered By | |-----------|-------------| | WRITE_CODE | Writing new functions, classes, modules | | MODIFY_CODE | Fixing, changing, refactoring existing code | | REVIEW_CODE | Reviewing PRs, diffs, or code | | DESIGN_SYSTEM | Designing architecture, APIs, databases | | WRITE_TESTS | Writing or adding tests | | DEPLOY_RELEASE | Deploying, releasing, CI/CD work | | DATA_WORK | Schema, migrations, queries, pipelines | | PLAN_FEATURE | Planning, scoping, estimating | | DESIGN_UI | UI/UX design and component work | | INCIDENT_RESPONSE | Outages, incidents, production issues |

Example checklist output:

### Pre-Delivery Checklist: WRITE_CODE

- [x] ENGINEERING: Names reveal intent, domain language used
- [x] ENGINEERING: Functions do one thing, ≤3 params
- [x] ARCHITECTURE: Dependencies point inward
- [x] TESTING: Tests written for new behavior
- [ ] SECURITY: **ACTION NEEDED** — User input passed to processOrder()
      without validation. Add input validation before processing.
- [x] DEVOPS: Configuration externalized

**Result: 5/6 passed. 1 action item.**

Team Mode

For comprehensive reviews, Claude launches parallel domain-specialist subagents:

> "Review this PR with the full team"

Claude (as Tech Lead) will:

1. Identify relevant domains
2. Launch specialist subagents in parallel (Software Engineer, QA, Security, etc.)
3. Each specialist analyzes from their domain's perspective
4. Results synthesized into a unified Team Review Report

Selective team launch matches specialists to the work type:

• Code PR → Software Engineer + QA + Security
• System design → Architect + Security + DevOps + Data + Product
• Feature planning → Product + UX + Delivery + Architect
• Incident → DevOps + Security + Data + Eng Manager

Before & After

| Prompt | Without | With required-reading | |--------|---------|----------------------| | Write a function | Vague names, too many params | Intent-revealing names, ≤3 params, SRP, CQS, checklist | | Design a system | Reasonable architecture | Quality attributes, threat model, SLOs, data model, ADRs, checklist | | Fix a bug | Fixes the bug | Fixes it + flags violations + runs MODIFY_CODE checklist | | Review code | Generic feedback | 10-domain systematic review with severity ratings | | Write tests | Basic assertions | TDD, pyramid, boundary values, proper test doubles | | Deploy | Manual steps | Zero-downtime strategy, rollback plan, observability check |

Pragmatism

required-reading is opinionated, not dogmatic.

• Scope-appropriate — won't apply system-level concerns to a utility function
• Prototype-aware — explicitly labeled throwaway code gets looser enforcement
• Performance-conscious — acknowledges that hot paths may trade readability for speed
• Legacy-friendly — won't demand a full rewrite during incremental improvement
• Noise-free — checklists surface real issues, not bureaucratic checkboxes

To bypass: "Skip enforcement" or "Quick and dirty mode." Claude will comply but note what it would change for production.

Project Structure

required-reading/
├── .claude-plugin/
│   ├── marketplace.json
│   └── plugin.json
├── bin/
│   └── install.js
├── skills/
│   ├── required-reading/
│   │   └── SKILL.md                          (core — 10 domains, checklists, team mode)
│   ├── required-reading-software-engineering/
│   │   └── SKILL.md                          (deep-dive specialist)
│   ├── required-reading-architecture/
│   │   └── SKILL.md
│   ├── required-reading-testing/
│   │   └── SKILL.md
│   ├── required-reading-security/
│   │   └── SKILL.md
│   ├── required-reading-devops/
│   │   └── SKILL.md
│   ├── required-reading-data-engineering/
│   │   └── SKILL.md
│   ├── required-reading-delivery/
│   │   └── SKILL.md
│   ├── required-reading-product/
│   │   └── SKILL.md
│   ├── required-reading-ux/
│   │   └── SKILL.md
│   └── required-reading-leadership/
│       └── SKILL.md
├── claude-md-snippet.md
├── package.json
└── README.md

License

MIT