<2-3 sentences summarising what the project addresses and solves, the target audience, and the benefit.>

This methodology provides a structured framework for collaborative software development with AI, addressing common issues like code bloat and architectural drift. It targets engineers and power users seeking to improve AI-generated code quality, maintainability, and consistency through systematic constraints and validation. The primary benefit is a more disciplined and empirically-driven development process, reducing debugging time and enhancing project reliability.

How It Works

The approach employs a four-stage process: AI Configuration (setting custom instructions), Collaborative Planning (structuring the project with AI), Systematic Implementation (focused, modular development with strict constraints like 150-line files), and Data-Driven Iteration (using benchmarks for optimization). This methodology leverages empirical data and systematic constraints to manage AI context effectively, breaking down complex tasks into manageable, focused questions that AI handles more reliably than broad requests.

Quick Start & Requirements

This is a methodology, not a software package with a direct installation command. Key steps involve configuring AI models with custom instructions (e.g., AI-PREFERENCES.XML), sharing project planning documents (METHODOLOGY.XML), and utilizing provided scripts like scripts/project_extract.py for context management and compliance checks. Users require access to AI models and a foundational understanding of software architecture principles. Links to specific project examples and detailed LLM evaluations are mentioned but not directly provided in the text.

Highlighted Details

• Enforces a strict 150-line limit per file to promote modularity, focus, and easier debugging.
• Demonstrates application through example projects like a Discord Bot Template, PhiCode Runtime, and PhiPipe CI/CD system.
• Includes a comparative evaluation of AI models (Grok 3, Claude Sonnet 4, DeepSeek-V3, Gemini 2.5 Flash) based on their adherence to the methodology.
• Features a project_extract.py tool for generating structured code snapshots, tracking compliance, and managing AI context.

Maintenance & Community

The README does not detail specific community channels (like Discord or Slack), contributor information beyond the author (Jay Baleine), or a public roadmap.

Licensing & Compatibility

The methodology is licensed under CC BY-SA 4.0 (Creative Commons Attribution-ShareAlike 4.0 International). This copyleft license requires derivative works to be shared under the same terms, which may impose compatibility considerations for commercial or closed-source projects.

Limitations & Caveats

AI models may still require occasional reminders to adhere to principles, and the author notes they cannot accurately assess the learning curve due to personal neurological differences. The methodology is not intended for simple scripts or users averse to planning, and it requires users to possess fundamental programming knowledge.