What is actually in the SKILL.md
A skill is only as good as its instructions, so we simply describe what is there, without making it sound better than it is. At the top sits a frontmatter: the name ai-scamper-innovation-coach, version 1.0.0 and MIT licence. Below that is a long description that determines when Claude picks up the skill on its own. Not only at the word SCAMPER, but also at brainstorming, product innovation, improving an idea, coming up with new product variants, redesigning a service, business model variations, how do I renew my product, what could be better about this, think out of the box, ideation session and help me innovate. Say something like that to Claude while the skill is loaded, and it switches on by itself.
SCAMPER was developed by Bob Eberle in 1971, building on Alex Osborn’s work systematising and popularising brainstorming. Its seven lenses prompt alternative directions, not guaranteed discoveries. The file gives each lens questions and now includes a complete illustrative set for the pizzeria.
The execution part consists of four steps. Step 1, understand the object. If it is not clear exactly what is being analysed, the skill first asks: what is the core product or core service, who is the current target audience, what is the current price or cost model, and what is the biggest pain point or limitation right now. There is a limit built straight into this that you rarely see in prompts: ask a maximum of three questions at once and then move on. So you get no questionnaire, at most a short round of questions.
Step 2, apply all seven lenses. Each letter is worked through systematically with three to five concrete, actionable ideas. Even the format is fixed: per lens the letter with its name and an English explanation where useful, then the central question tailored to your object, then numbered ideas where each idea is followed by a short explanation of why it may help or which problem it aims to solve. No vague statements, only concrete proposals: that is what it literally says.
Step 3, select the top three. After the seven lenses, the skill picks the three strongest ideas from all the variants, on three criteria: high impact on the user's main KPI such as revenue, retention or margin, realistic feasibility within 90 days, and differentiation from competitors. For each chosen idea it delivers four things: which SCAMPER lens it is, the idea in one sentence, the first three action steps to test it, and which assumption needs to be validated before you move on.
Step 4, suggest a combination. The file closes with a combination of two letters as another hypothesis to test, not proof that stacking produces the strongest innovation. In the pizzeria example those are Eliminate and Substitute, combined into a subscription with three options that change weekly.
Finally there is a Tone and rules block with six points, four of which explicitly forbid something. Clear language and no jargon. No emoji bullets, just normal bullet points. Concrete and businesslike, not woolly. Show that the object is genuinely understood by echoing specific words from your input. Challenge the assumption and do not flatter. And: if an idea is stupid or risky, say so. That last point is the most interesting one, because it stops the skill from being a yes-man.