That a Skill is a prompt is easy to see: human experience, judgment standards, and procedures, compressed into a template you can invoke over and over.
Design is a prompt too. Every revision round tells the machine what counts as right, what counts as good-looking.
Law is a prompt. Statutes are instructions written for judges, lawyers, enforcers — the execution models of society. One sentence decides how millions act. Law is the System Prompt written for humanity’s giant model.
Once you see this, look back at your own trade: all knowledge work seems to be prompt engineering.
Which makes a tempting idea: treat prompts as the core asset. Is a media business’s real moat just a few good prompts?
No. A prompt is a soft asset with no legal wall. Worse: given your output, your prompt can be reverse-engineered. The better you write it, the easier it is to steal. I hold the same worry about my own tools — if Tencent wanted one, it could ship in a day. A prompt is even cheaper to take.
Someone will say: then compete on writing better prompts. That road has no exit. You use prompts, so does everyone; you upgrade yours into Skills, so does everyone. When technique converges, everyone is the same.
The advantage isn’t in the prompt. It’s in the context behind it. Take Xiaohongshu. “Family, who understands” — that’s the shallowest layer. One level down: the mechanics of stirring emotion, triggering purchases, displaying the life people wish they had. One level deeper: keyword libraries, corpora, vector stores — words as one layer, emotion as another. Stack the layers, run the meta-analysis, and you’re holding a workflow database nobody else has.
How do human and machine divide the work at that point? In parallel, on the same layer. The human owns brand direction; the machine owns analysis. Two jobs, neither replaceable by the other.
Other people see the post you published. You see the workflow database behind it. One is the result, the other is the process. Anyone can write a prompt. Nobody can copy your process.
