Business Idea Validator AI Skill — AI skill for Claude Code

Business Idea Validator — a Claude Skill

Tell it a business idea. It researches the market, scores the idea against a computed rubric, and hands you a brief that opens with the whole decision on one page:

Decision
  VERDICT           Weak · 42.0/100
  WHY NOW           Long-context inference crossed viability in 2025;
                    the same product in 2022 cost 40x more per document.
  REFERENCE CLASS   Vertical AI tools sold to regulated SMBs under $200/mo.
                    Roughly 1 in 12 reach $1M ARR. Founder projects $4M by year two.
  BIGGEST RISK      No one has been asked to pay; all 24 interviews ended
                    in encouragement.
  NEXT ACTION       Run 20 choice-based WTP interviews within two weeks.
  STOP IF           Fewer than 8 of 30 name this a top-3 problem by 22 Sep.

npx business-idea-validator

Then describe an idea, or use `/business-idea-validator`.

Why the verdict is a number

Ask any model what it thinks of your idea and it will tell you the idea is great. Asking it to be honest does not fix that — instructions about tone do not change where a judgment lands. A rubric does.

Six weighted components produce a score, the score maps to a band, and the model gets exactly one band of override which it has to print and justify. It also says what would make the idea work, not only why it might not.

What you get

Every evaluation produces a brief — as a shareable page, a PDF, or HTML — alongside a machine-readable `verdict.json`.

**Abridged example output** (bootstrap-path, nascent). Component weights are set by the capital path — moat is the whole thesis for a venture play and nearly irrelevant for a niche tool whose defence is that the niche is too small to be worth attacking:

Score                                          raw    of  weight   score
  AI-era check      3 N/A (no model in product) 12    20      25    15.0
  Demand evidence   2 paying, 6 verbal commits  16    20      30    24.0