Prompt once. Verify everything. Loop until the goal is achieved.
AI Engineering Operating System is a public methodology and documentation framework for turning AI coding agents into disciplined software engineering systems.
It defines reusable SDLC loops, Mermaid diagrams, verification gates, model routing, memory rules, human review checkpoints, and continuous improvement workflows for local-first AI development.
Most AI coding workflows still behave like ad-hoc chats. AI-OS turns that into an auditable engineering process:
- set a clear goal
- fan out to specialist roles
- produce a plan
- review risk before broad changes
- implement in small reversible steps
- verify using strong gates
- update docs, memory, wiki, and roadmap
- repeat until no valuable improvement remains
- Documentation map
- Quickstart
- Master operating manual
- Continuous improvement loop
- Definition of Done
- Maturity model
- Mermaid diagram catalog
- Loop catalog
- Verification gates
- Prompt templates
- ChatGPT web operating prompt
- Evaluation framework
- Completeness checklist
- Readiness status
- Wiki home
Execute AI Engineering Operating System from this repository.
Task:
[YOUR TASK HERE]
Load context first.
Select the right loop.
Fan out analysis to specialist roles.
Create a plan before implementation.
Review risk before broad changes.
Implement work in small reversible steps.
Verify before declaring completion.
Return to the earliest failing phase when verification fails.
Update docs and memory.
Stop only when the Definition of Done passes.
This repository documents the instructions, diagrams, and operating model I use when working with AI coding agents and ChatGPT web sessions.
The goal is to make AI-assisted development visible, repeatable, auditable, and production-oriented.
AI-OS targets Level 5 of its maturity model: reusable prompts, task-specific loops, verifier-driven completion, governance controls, and controlled continuous improvement.