A Skill Nobody Uses Is Just a File in a Folder
Module 5 of the Freedom Ag training shipped five Claude skills: humanizer, spin-selling, influence, storybrand, and pdf. We've written about each one. What we haven't written about is the five prompts that came with them.
Not the self-teaching prompt that makes Claude explain a skill from the file itself. The scenario prompts: one per skill, each one placing the skill in a concrete situation from the client's actual week.
Those prompts are the difference between a skill that gets installed and a skill that gets used.
What the Scenario Prompts Do
Each one opens the same way: "Use the [skill name] skill..." That line activates the skill by name. Everything after it is a specific scenario from ag input sales.
The humanizer prompt doesn't say "rewrite this email." It says: before you rewrite anything, name three specific things that make it read as AI-written, and quote the exact lines. Then give me one rewrite. That sequence forces Claude to diagnose before it touches the draft. If the diagnosis is wrong, you catch it before the rewrite.
The spin-selling prompt doesn't say "help me sell." It interviews you first: who am I calling, how many acres, what we've talked about before, what price objection they threw last time. Four questions, numbered, then it stops and waits. You don't get a call plan until you've given Claude something real to work with.
The influence prompt asks you to paste a seasonal push you're about to send. It then evaluates each of Cialdini's seven principles against it and rates how honest each use is. The filter that matters: scarcity and authority, which are the two most commonly manufactured. It flags them plainly.
The pdf prompt, before doing anything with the files, reports back exactly what it found: page count, what was on each page, and anything that didn't extract cleanly. Numbers that were blurry or ambiguous get flagged. You know what you're working with before you ask a question.
Why This Structure
Every one of them gates on something real before producing anything. Read the file first. Ask the four questions first. Quote the actual lines first. Report what you found before interpreting it.
That's not a stylistic choice. Skills encode frameworks built for careful analysis: a sales methodology, a persuasion checklist, a narrative structure. Those frameworks produce garbage if the input is vague. The scenario prompt forces specificity before the skill can do its job.
The Scenario Is the Teaching
You can read the README for the spin-selling skill. It will tell you it's based on Rackham's research and that it's designed for complex B2B sales with long cycles. You'll understand what it does in the abstract.
Running the scenario prompt teaches you something the README can't: what it feels like when Claude won't let you skip the discovery questions. When it asks who you're calling on before it gives you a single talking point. When it pushes back because you said "a big grower" instead of naming the number of acres and the crop.
That's the behavior the skill produces. The scenario prompt is how you see it before you need it in a real call.
What We Ship Now
For any skills pack we build, we now treat the scenario prompts as part of the deliverable, not optional documentation. One per skill. Each one paired to a pressure point the client actually faces. Each one gated on something real before the output appears.
A skill without a scenario prompt is asking the client to trust a README. A skill with one gives them a result they can evaluate on day one.