⚡OAIESv1.0.0
🚀 The AI Engineering Standard

Open AI Engineering
Standard

The production-ready, enterprise-grade foundation for AI systems. One standard. One stack. No exceptions.

Read the Standard →AI SDLC PromptsExplore the Standards Portal
10
Handbook chapters
12
Patterns
15
SDLC prompts
11
Maturity levels

11-Level Maturity Model

Enter at your current level. Each level has a readiness gate before progression.

🧱
Level 0
AI Foundations
The Harness Principle, failure modes, core mental models
✍️
Level 1
Prompt Engineering
XML-structured prompts, anti-patterns, system prompt templates
📦
Level 2
Context Engineering
Context fidelity, 13 context types, progressive disclosure
🛠️
Level 3
Skill Engineering
Reusable procedural knowledge for debug, review, security
🤖
Level 4
Agent Engineering
OAIES agent spec, hooks, subagents, termination criteria
🕸️
Level 5
Multi-Agent Systems
Supervisor, pipeline, parallel patterns, communication protocol
🧠
Level 6
Memory & Knowledge
Hybrid RAG, vector databases, episodic and semantic memory
⚙️
Level 7
LLMOps
Evaluation gates, PromptOps, cost optimization, observability
🔄
Level 8
AI SDLC
Stage workflow with 15 production prompts and quality gates
🏢
Level 9
Enterprise AI
Governance, EU AI Act, provider-neutral architecture, audit logging
📖
Level 10
AI Org Playbook
Roles, operating models, hiring criteria, culture principles

What Makes OAIES Different

Most AI repos say “here are 3 ways.” OAIES says: this is the one way. This is why.

📘
Practitioner Handbook
Skills vs prompts, context packages, planning gates, HITL, PR review, evals, hooks, MCP, and token discipline — written for teams that ship UI with AI.
🎯
12 AI Engineering Patterns
Operational envelopes (planner, reviewer, release, hotfix…). Open the pattern README from the Patterns menu — not a dump of stub components.
🔒
Security-First by Default
Security skill, review gates, and enterprise control maps. High-impact effects require human approval.
📊
Evaluation CI/CD Gate
GitHub Actions workflow that blocks PRs when LLM evaluation tests fail. Prompts are treated like code.
🔄
AI SDLC + 15 production prompts
Stage workflow with versioned prompts for story, plan, coding, security, incident, RCA, and knowledge capture.
🍳
Technology cookbooks
React, Next.js, microfrontends, and more — native commands, failure classes, and rollback — linked from curated nav.

Ready to raise your AI engineering standard?

Start with Level 0: The Harness Principle. Every production AI system starts here.

The Harness Principle →Browse Patterns