Agentic AI and Skills: I think this is where things are going
Lately, I have been thinking about agentic AI in general and skills in particular.
To me, agentic AI is not just a chatbot giving answers. It is more like an AI system that can take a goal, break it into steps, use tools, check the result, and keep moving forward.
But the part I find interesting is skills.
I don’t see skills as just prompts. I see them more like reusable playbooks. Almost like a way to teach the AI how you want certain work done.
For example, instead of explaining the same deployment process over and over again, you could have a deployment skill. Instead of explaining your code review process every time, you could have a code review skill. Same for testing, documentation, database setup, security checks, or app generation.
So the agent becomes the one deciding what needs to happen, and the skills become the repeatable ways of doing the work.
That feels like a big shift to me.
We are moving from:
“AI, answer this question”
to:
“AI, use the right process and get this done.”
I think this is especially important for software development, because a lot of value is not only in generating code, but in repeating the right process consistently: build, test, deploy, validate, improve.
Are skills just better prompts, or are they becoming something closer to reusable operating procedures for AI agents?
I am genuinely curious what everyone thinks about the subject. Let’s explore together.
Additionally, are you using tools like OpenClaw or Hermes?