It will help you improve your skills and understand how to start this journey!
In this article I’m sharing how I’m increasingly relying on AI in my daily work—and even to contribute to OSS projects.
AI models are trained on what’s publicly available. They handle React, Django, and Kubernetes just fine. But ask them about a niche open-source tool you actually use at work, and they struggle. Not because the tool is hard, but because there’s barely any training data about it.
That’s the problem skill.md solves. It’s a simple format that lets you write specific instructions for any task or technology, and feed them to an AI. Instead of guessing, the model follows your boundaries, your steps, and your reference material.
I use AI to write these skill.md files, then reuse them across projects. The difference is noticeable: the AI stops making things up and starts giving answers that actually fit the stack I’m working with.
I’ve been publishing what I’ve built at https://github.com/CodeAtCode/oss-ai-skills. Each file defines a scope, provides working examples, and links to real documentation so the AI can point you to the right resources instead of inventing them.
Browse it, copy what’s useful, and add your own. The more skill.md files people share, the less we all have to deal with AI guessing its way through unfamiliar tools.
