Engineering Standards
A thorough, cross-domain software engineering standard for AI coding agents. Works with Claude Code, Codex, Cursor, Gemini CLI, Aider, and any agent that reads an instructions file.
It encodes the canon of software engineering — SOLID, Clean Code, Designing Data-Intensive Applications, A Philosophy of Software Design, 97 Things Every Software Architect Should Know, Release It!, 12-Factor, OWASP, CAP, and battle-tested practice — as enforceable, language-agnostic rules the agent applies silently to everything it writes, reviews, and designs.
One lean always-on core (the universal laws) plus on-demand deep-dives for each domain — so it stays lightweight no matter how large the standard grows.
What it covers
| Domain | Reference file |
|---|
| System design, tradeoffs, quality attributes, DDD, ADRs, technical debt | architecture |
| OOP / FP / procedural, immutability, pure functions, functional core | paradigms |
| Complexity analysis, data-structure choice, time/space tradeoffs | algorithms |
| Services, APIs, error handling, jobs, caching, rate limiting | backend |
| Schemas, transactions, locking, indexing, replication, store choice | data |
| Threads, async, races, memory models, backpressure, distributed coordination | concurrency |
| UI structure, state, rendering strategy, accessibility, Core Web Vitals | frontend |
| Flutter / Android / iOS, offline-first, lifecycle, release & signing | mobile |
| Deployment, scaling, resilience, observability, IaC, cost, multi-region | cloud |
| Memory, allocation, syscalls, cache, lock-free, NUMA, I/O models | systems |
| Input boundaries, authz, crypto, threat modeling, OWASP Top 10 | security |
| Test strategy, determinism, test doubles, property/load/chaos testing | testing |
| Commits, branching, pull requests, code review, continuous integration | git-craft |
| LLM/AI feature design, prompt engineering, evals, RAG, prompt injection, agents, cost/token management | ai-engineering |
The full standard lives in AGENTS.md (the universal laws, agent-neutral) plus the 14 domain files under skills/engineering-standards/reference/.
Install
Choose your installation method below. You can install the engineering standards locally into a specific project root, or globally to automatically configure all your AI agents without cloning the repository.
🚀 A. Unix Quick Installer (macOS & Linux)
Run the automated shell installer via curl:
# Local (Installs standards to the current directory)
curl -fsSL https://raw.githubusercontent.com/S13G/engineering-standards-plugin/main/install.sh | sh
# Global (Installs to ~/.engineering-standards and configures global profiles)
curl -fsSL https://raw.githubusercontent.com/S13G/engineering-standards-plugin/main/install.sh | sh -s -- --global
💻 B. Windows PowerShell Installer (Windows)
Run the automated PowerShell installer via irm (Invoke-RestMethod):
# Local (Installs standards to the current directory)
irm https://raw.githubusercontent.com/S13G/engineering-standards-plugin/main/install.ps1 | iex
# Global (Installs to ~/.engineering-standards and configures global profiles)
irm https://raw.githubusercontent.com/S13G/engineering-standards-plugin/main/install.ps1 | iex -Arguments "-Global"
📦 C. Node.js & npm/npx Installer (Cross-platform)
You can run the installer instantly with npx, or install it globally with npm:
Option 1: Run instantly using npx
# Local (Installs standards to the current directory)
npx engineering-standards-plugin
# Global (Installs to ~/.engineering-standards and configures global profiles)
npx engineering-standards-plugin --global
Option 2: Install globally using npm
# Install globally on your system
npm install -g engineering-standards-plugin
# Run the installer in any project directory
engineering-standards
# Run globally to configure system-wide agent profiles
engineering-standards --global
Agent Compatibility Matrix
Once installed, standard files map directly to agent inputs:
| Agent | Entry point | Setup |
|---|
| Codex | AGENTS.md | Read automatically. Just have AGENTS.md + skills/ in the project. |
| Cursor | .cursorrules | Read automatically once present in the project root. |
| Gemini CLI | GEMINI.md | Read automatically; it points to AGENTS.md. |
| Aider | AGENTS.md | Run with aider --read AGENTS.md. |
| Claude Code | plugin (recommended) or CLAUDE.md | See below — the plugin adds automatic, progressive loading. |
| Any other | AGENTS.md | Include it in the agent's context and ensure it can open reference/*.md on demand. |
