AI Policy
The availabilty of Large Language Models for general use has had an undeniable impact on the state of software development over the past few years. Specialized and “everything” models alike can generate massive amounts of code for almost any program imaginable. Generating a website like my own or its projects would not be a complex feat using the latest AI models and specific enough prompts.
Professional AI Use
I believe it is important for today's engineers to have a thorough understanding of the strengths and limitations of these tools. Maintaining code that can be understood and manipulated by both humans and AI is now a core responsibility of engineers at enterprise organizations. As such, I create and use these tools daily on the job, including...
- Custom agents and subagents
- Agent skills
- MCP servers
- AI Frameworks (e.g. Research — Plan — Implement )
- Project-specific context for LLMs
- AI-driven PR reviews
Personal AI Use
My site and its projects have different goals than the software at enterprise organizations. I use the act of building them to expand my understanding of coding techniques and technologies that I want to explore. The projects are ones I get excited about programming directly instead of handing them off to a web of agents. As such, I am very selective of how I use AI outside of work.
As of now, I've only used the following AI tools while developing the projects shared on this site:
- Code completion suggestions in IDE via GitHub Copilot
- Chats with AI chatbots such as ChatGPT/Claude
I did not use agents to generate code for these projects. Any chats that generated code snippets were used selectively with a complete understanding of the solution offered. I never copied a complete file/group of files generated by a chat for my own use in these projects.
That being said, I used AI agents to build parts of shanemcfadden.dev. I anticipate that I will begin using agentic engineering techniques for some future personal projects to better understand how far I can push this technology in a meaningful direction.
Should I begin using agents to develop the projects on this site, I will add indicators to each project page outlining how I used AI to build/maintain each project. For now, safely assume that all project source code was written directly by me.